Skip to content

Browse by technique

Mentions in full text

Papers ordered by how often the term occurs (count in brackets). Only papers with a local PDF are counted.

Bootstrapping (208 papers)

Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (231), General Functional Bootstrapping using CKKS (226), The Beginner's Textbook for Fully Homomorphic Encryption (214), META-BTS: Bootstrapping Precision Beyond the Limit (214), Bootstrapping Bits with CKKS (173), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (151), Bootstrapping GBFV with CKKS (150), SHIP: A Shallow and Highly Parallelizable CKKS Bootstrapping Algorithm (143), High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (143), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (138), TFHE: Fast Fully Homomorphic Encryption over the Torus (137), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (134), Secure numerical simulations using fully homomorphic encryption (127), Demystifying bootstrapping in fully homomorphic encryption (118), Fully Homomorphic Encryption for Machine Learning (115), Leveraging Discrete CKKS to Bootstrap in High Precision (113), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (106), Efficient Homomorphic Integer Computer from CKKS (105), PaCo: Bootstrapping for CKKS via Partial CoeffToSlot (103), Bootstrapping for BGV and BFV Revisited (102), High-Precision Exact FHE Made Simple, General, and Fast (100), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (98), Large-Plaintext Functional Bootstrapping in FHE with Small Bootstrapping Keys (95), FINAL: Faster FHE Instantiated with NTRU and LWE (93), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (92), Fully Homomorphic Encryption for Cyclotomic Prime Moduli (85), (Leveled) Fully Homomorphic Encryption without Bootstrapping (85), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (84), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (84), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (81), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (81), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (80), Low-Depth Bootstrapping for Matrix-Native FHE (77), Plaintext-Ciphertext Matrix Multiplication and FHE Bootstrapping: Fast and Fused (72), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (71), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (69), Efficient FHEW Bootstrapping with Small Evaluation Keys, and Applications to Threshold Homomorphic Encryption (67), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (67), Faster Logical Operations from Discrete CKKS (65), OpenFHE: Open-Source Fully Homomorphic Encryption Library (64), Security guidelines for implementing homomorphic encryption (63), BASALISC: Programmable Hardware Accelerator for BGV Fully Homomorphic Encryption (63), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (62), High-Precision Homomorphic ALU over Arbitrary Moduli with O (1) Bootstrapping (61), Stay in Your Lane: Fast Arithmetic over Large Finite Rings with CKKS (60), Better Bootstrapping for Approximate Homomorphic Encryption (60), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (59), Bootstrapping for Approximate Homomorphic Encryption (59), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (59), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (58), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (57), SoK: Private LLM Inference using Approximate Homomorphic Encryption (57), LFHE: Fully Homomorphic Encryption with Bootstrapping Key Size Less than a Megabyte (52), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (51), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (51), CKKS Bootstrapping in less than 10ms (48), Improved Bootstrapping for Approximate Homomorphic Encryption (48), FINALLY: A Multi-Key FHE Scheme Based on NTRU and LWE (47), Faster bootstrapping with polynomial error (47), Multi-Key Homomorphic Encryption from TFHE (46) …

Rotations (144 papers)

The Beginner's Textbook for Fully Homomorphic Encryption (120), Efficient FHEW Bootstrapping with Small Evaluation Keys, and Applications to Threshold Homomorphic Encryption (99), CKKS Bootstrapping in less than 10ms (81), TRICYCLE: Private transformer inference with tricyclic encodings (72), SHIP: A Shallow and Highly Parallelizable CKKS Bootstrapping Algorithm (66), STIP: Efficient and secure non-interactive transformer inference via compact packing (66), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (65), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (64), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (63), Algorithms in HElib (60), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (54), THEMIS: A Co-Designed System for Encrypted Transformer Inference (53), ARION: Attention-optimized transformer inference on encrypted data (50), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (49), Secure numerical simulations using fully homomorphic encryption (43), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (42), THOR: Secure transformer inference with homomorphic encryption (40), LFHE: Fully Homomorphic Encryption with Bootstrapping Key Size Less than a Megabyte (39), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (39), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (38), Secure outsourced matrix computation and application to neural networks (33), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (32), Low-Depth Bootstrapping for Matrix-Native FHE (32), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (31), PaCo: Bootstrapping for CKKS via Partial CoeffToSlot (31), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (30), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (30), Secure human action recognition by encrypted neural network inference (28), Fully Homomorphic Encryption for Matrix Arithmetic (28), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (28), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (23), Optimized Privacy-Preserving CNN Inference With Fully Homomorphic Encryption (21), Bootstrapping for Approximate Homomorphic Encryption (19), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (19), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (19), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (19), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (19), SoK: Private transformer-based model inference (18), Pasta: A Case for Hybrid Homomorphic Encryption (18), EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computation (17), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (17), TFHE: Fast Fully Homomorphic Encryption over the Torus (17), Privft: Private and fast text classification with homomorphic encryption (16), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (16), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (16), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (16), SoK: Private LLM Inference using Approximate Homomorphic Encryption (16), High-Precision Homomorphic ALU over Arbitrary Moduli with O (1) Bootstrapping (15), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (14), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (14), CrypTFlow2: Practical 2-Party Secure Inference (14), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (14), Secure transformer inference made non-interactive (14), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (13), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (13), Fast and Accurate Homomorphic Softmax Evaluation (13), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (11), Multi-key Fully Homomorphic Encryption with Non-Interactive Setup in the Plain Model (10), Improved Bootstrapping for Approximate Homomorphic Encryption (10), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (10) …

SIMD packing (141 papers)

Private machine learning classification based on fully homomorphic encryption (74), Fully Homomorphic SIMD Operations (67), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (50), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (39), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (36), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (35), BumbleBee: Secure Two-party Inference Framework for Large Transformers (31), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (27), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (27), Secure transformer inference made non-interactive (25), Secure outsourced matrix computation and application to neural networks (21), High-Precision Homomorphic ALU over Arbitrary Moduli with O (1) Bootstrapping (21), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (17), Algorithms in HElib (16), Bootstrapping Bits with CKKS (14), Secure human action recognition by encrypted neural network inference (13), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (13), BLEACH: Cleaning Errors in Discrete Computations over CKKS (13), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (13), High-Precision Exact FHE Made Simple, General, and Fast (12), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (12), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (12), Fully Homomorphic Encryption for Cyclotomic Prime Moduli (11), CipherGPT: Secure Two-Party GPT Inference (11), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (11), SoK: Private LLM Inference using Approximate Homomorphic Encryption (11), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (10), Efficient Homomorphic Integer Computer from CKKS (10), A comprehensive survey on secure outsourced computation and its applications (9), Faster Logical Operations from Discrete CKKS (9), THOR: Secure transformer inference with homomorphic encryption (9), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (8), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (8), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (8), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (8), SoK: Private transformer-based model inference (7), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (7), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (7), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (7), Multiparty Computation from Somewhat Homomorphic Encryption (6), Fully Homomorphic Encryption on the Ring of Gaussian Periods (6), Bootstrapping GBFV with CKKS (6), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (6), CKKS Bootstrapping in less than 10ms (6), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (6), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (6), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (6), OpenFHE: Open-Source Fully Homomorphic Encryption Library (6), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (6), ARION: Attention-optimized transformer inference on encrypted data (6), Demystifying bootstrapping in fully homomorphic encryption (5), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (5), FheFL: Fully Homomorphic Encryption Friendly Privacy-Preserving Federated Learning with Byzantine Users (5), TFHE: Fast Fully Homomorphic Encryption over the Torus (5), Fully Homomorphic Encryption for Matrix Arithmetic (5), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (5), Leveraging Discrete CKKS to Bootstrap in High Precision (5), General Functional Bootstrapping using CKKS (5), TRICYCLE: Private transformer inference with tricyclic encodings (5), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (5) …

Convolution (138 papers)

Optimized Privacy-Preserving CNN Inference With Fully Homomorphic Encryption (135), Secure human action recognition by encrypted neural network inference (99), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (77), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (71), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (61), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (59), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (52), Slalom: Fast, Verifiable and Private Execution of Neural Networks in Trusted Hardware (49), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (35), CrypTFlow2: Practical 2-Party Secure Inference (34), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (31), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (31), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (29), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (26), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (25), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (22), TenSEAL: A Library for Encrypted Tensor Operations Using Homomorphic Encryption (22), Secure outsourced matrix computation and application to neural networks (20), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (20), Exploring Bitslicing Architectures for Enabling FHE-Assisted Machine Learning (20), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (17), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (17), Delphi: A Cryptographic Inference Service for Neural Networks (17), Attention is all you need (16), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (16), On Fully Homomorphic Encryption for Privacy-Preserving Deep Learning (15), Activate Me!: Designing Efficient Activation Functions for Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (13), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (12), Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering (12), CrypTen: Secure Multi-Party Computation Meets Machine Learning (12), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (11), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (11), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (11), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (11), Homomorphic Encryption for Machine Learning and Artificial Intelligence Applications (10), Fully Homomorphically Encrypted Deep Learning as a Service (9), Touch of Privacy: A Homomorphic Encryption-Powered Deep Learning Framework for Fingerprint Authentication (9), fairseq: A fast, extensible toolkit for sequence modeling (8), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (7), Cheddar: A Swift Fully Homomorphic Encryption Library Designed for GPU Architectures (7), ABY3: A Mixed Protocol Framework for Machine Learning (6), Fully Homomorphic Encryption for Machine Learning (6), Secure federated learning applied to medical imaging with fully homomorphic encryption (6), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (6), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (6), TensorFHE: Achieving Practical Computation on Encrypted Data Using GPGPU (6), zkLLM: Zero Knowledge Proofs for Large Language Models (6), Programmable Bootstrapping Enables Efficient Homomorphic Inference of Deep Neural Networks (6), Converting transformers to polynomial form for secure inference over homomorphic encryption (6), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (5), Private pathological assessment via machine learning and homomorphic encryption (5), Low-Depth Bootstrapping for Matrix-Native FHE (5), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (5), Theory, analysis, and best practices for sigmoid selfattention (4), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (4), Fully Homomorphic Encryption with Applications to Privacy-Preserving Machine Learning (4), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (4), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (4), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (4), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (4) …

Rescaling (122 papers)

Approximate Homomorphic Encryption with Reduced Approximation Error (117), EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computation (109), The Beginner's Textbook for Fully Homomorphic Encryption (108), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (96), Cheddar: A Swift Fully Homomorphic Encryption Library Designed for GPU Architectures (64), On the precision loss in approximate homomorphic encryption (29), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (27), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (23), META-BTS: Bootstrapping Precision Beyond the Limit (23), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (23), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (20), Leveraging Discrete CKKS to Bootstrap in High Precision (20), Homomorphic Encryption for Arithmetic of Approximate Numbers (20), Plaintext-Ciphertext Matrix Multiplication and FHE Bootstrapping: Fast and Fused (19), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (17), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (17), SHIP: A Shallow and Highly Parallelizable CKKS Bootstrapping Algorithm (17), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (17), Stay in Your Lane: Fast Arithmetic over Large Finite Rings with CKKS (16), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (16), High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (15), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (15), Secure numerical simulations using fully homomorphic encryption (14), TensorFHE: Achieving Practical Computation on Encrypted Data Using GPGPU (14), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (14), A Full RNS Variant of Approximate Homomorphic Encryption (14), ModHE: Modular Homomorphic Encryption Using Module Lattices: Potentials and Limitations (13), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (13), Bootstrapping GBFV with CKKS (12), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (12), FIDESlib: A Fully-Fledged Open-Source FHE Library for Efficient CKKS on GPUs (12), Low-Depth Bootstrapping for Matrix-Native FHE (12), Privft: Private and fast text classification with homomorphic encryption (11), General Functional Bootstrapping using CKKS (11), OpenFHE: Open-Source Fully Homomorphic Encryption Library (10), Approximate Homomorphic Encryption over the Conjugate-invariant Ring (10), Transciphering Framework for Approximate Homomorphic Encryption (9), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (9), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (8), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (8), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (8), Bootstrapping Bits with CKKS (8), On the Security of Homomorphic Encryption on Approximate Numbers (8), SPIRAL: Fast, high-rate single-server PIR via FHE composition (7), zkLLM: Zero Knowledge Proofs for Large Language Models (7), Fully Homomorphic Encryption for Matrix Arithmetic (7), Practical secure inference algorithm for a fine-tuned large language model based on fully homomorphic encryption (7), High-Precision Homomorphic ALU over Arbitrary Moduli with O (1) Bootstrapping (6), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (6), HEonGPU: a GPU-based Fully Homomorphic Encryption Library 1.0 (6), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (6), TFHE: Fast Fully Homomorphic Encryption over the Torus (6), Bootstrapping for Approximate Homomorphic Encryption (6), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (5), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (5), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (5), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (5), TREBUCHET: Fully Homomorphic Encryption Accelerator for Deep Computation (5), PaCo: Bootstrapping for CKKS via Partial CoeffToSlot (5), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (5) …

Key switching (116 papers)

Revisiting Homomorphic Encryption Schemes for Finite Fields (84), Accelerating HE Operations from Key Decomposition Technique (77), The Beginner's Textbook for Fully Homomorphic Encryption (67), Fully Homomorphic Encryption for Matrix Arithmetic (58), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (52), LFHE: Fully Homomorphic Encryption with Bootstrapping Key Size Less than a Megabyte (51), FINALLY: A Multi-Key FHE Scheme Based on NTRU and LWE (49), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (46), Approximate Homomorphic Encryption with Reduced Approximation Error (38), TFHE: Fast Fully Homomorphic Encryption over the Torus (35), BASALISC: Programmable Hardware Accelerator for BGV Fully Homomorphic Encryption (32), Multi-Key Homomorphic Encryption from TFHE (31), FINAL: Faster FHE Instantiated with NTRU and LWE (29), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (27), Stay in Your Lane: Fast Arithmetic over Large Finite Rings with CKKS (24), Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (23), High-Precision Exact FHE Made Simple, General, and Fast (22), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (20), Better Bootstrapping for Approximate Homomorphic Encryption (19), ModHE: Modular Homomorphic Encryption Using Module Lattices: Potentials and Limitations (18), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (18), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (18), Plaintext-Ciphertext Matrix Multiplication and FHE Bootstrapping: Fast and Fused (18), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (17), Multi-key Fully Homomorphic Encryption with Non-Interactive Setup in the Plain Model (16), Large-Plaintext Functional Bootstrapping in FHE with Small Bootstrapping Keys (16), OpenFHE: Open-Source Fully Homomorphic Encryption Library (16), TRICYCLE: Private transformer inference with tricyclic encodings (16), Ajax: Fast Threshold Fully Homomorphic Encryption without Noise Flooding (15), Fully Homomorphic Encryption for Machine Learning (15), Low-Depth Bootstrapping for Matrix-Native FHE (15), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (15), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (15), Secure numerical simulations using fully homomorphic encryption (14), Improved Security for a Ring-Based Fully Homomorphic Encryption Scheme (14), Bootstrapping for Approximate Homomorphic Encryption (14), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (14), THOR: Secure transformer inference with homomorphic encryption (14), Efficient FHEW Bootstrapping with Small Evaluation Keys, and Applications to Threshold Homomorphic Encryption (13), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (13), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (13), Security guidelines for implementing homomorphic encryption (12), SPIRAL: Fast, high-rate single-server PIR via FHE composition (12), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (12), F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (12), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (11), FHEW: Bootstrapping Homomorphic Encryption in Less Than a Second (11), Fully Homomorphic Encryption without Modulus Switching from Classical GapSVP (11), General Functional Bootstrapping using CKKS (11), Trace-Factored BigSwitch for Matrix-Friendly FHE (10), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (10), STIP: Efficient and secure non-interactive transformer inference via compact packing (10), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (10), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (9), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (9), HEonGPU: a GPU-based Fully Homomorphic Encryption Library 1.0 (9), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (9), On the precision loss in approximate homomorphic encryption (8), CKKS Bootstrapping in less than 10ms (8), A Survey on Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (8) …

Matrix multiplication (111 papers)

TRICYCLE: Private transformer inference with tricyclic encodings (131), Fully Homomorphic Encryption for Matrix Arithmetic (103), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (68), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (62), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (53), THOR: Secure transformer inference with homomorphic encryption (51), Secure outsourced matrix computation and application to neural networks (50), Encryption-friendly LLM architecture (48), Iron: Private Inference on Transformers (37), Low-Depth Bootstrapping for Matrix-Native FHE (34), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (32), ENSI: Efficient non-interactive secure inference for large language models (29), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (27), ARION: Attention-optimized transformer inference on encrypted data (26), SoK: Private LLM Inference using Approximate Homomorphic Encryption (25), Nimbus: Secure and Efficient Two-Party Inference for Transformers (24), Large language model inference acceleration: A comprehensive hardware perspective (23), Secure transformer inference made non-interactive (23), BumbleBee: Secure Two-party Inference Framework for Large Transformers (22), Plaintext-Ciphertext Matrix Multiplication and FHE Bootstrapping: Fast and Fused (21), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (21), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (20), Generalized BGV, BFV, and CKKS for Homomorphic Encryption over Matrix Rings (20), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (20), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (17), A comprehensive survey on secure outsourced computation and its applications (16), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (14), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (14), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (14), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (13), zkLLM: Zero Knowledge Proofs for Large Language Models (12), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (12), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (12), Pasta: A Case for Hybrid Homomorphic Encryption (11), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (11), Bitnet: Scaling 1-bit transformers for large language models (10), Machine Learning Training on Encrypted Data with TFHE (10), THEMIS: A Co-Designed System for Encrypted Transformer Inference (10), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (10), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (9), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (8), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (8), SIGMA: Secure GPT Inference with Function Secret Sharing (8), SoK: Private transformer-based model inference (7), TenSEAL: A Library for Encrypted Tensor Operations Using Homomorphic Encryption (7), CrypTen: Secure Multi-Party Computation Meets Machine Learning (7), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (7), CipherGPT: Secure Two-Party GPT Inference (7), CrypTFlow2: Practical 2-Party Secure Inference (7), The Beginner's Textbook for Fully Homomorphic Encryption (6), TensorFHE: Achieving Practical Computation on Encrypted Data Using GPGPU (6), Trace-Factored BigSwitch for Matrix-Friendly FHE (6), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (6), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (6), ABY3: A Mixed Protocol Framework for Machine Learning (5), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (5), MPCFormer: Fast, Performant and Private Transformer Inference with MPC (5), A scalable multi-GPU framework for encrypted large-model inference (5), One server for the price of two: Simple and fast single-server private information retrieval (4), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (4) …

RNS (109 papers)

The Beginner's Textbook for Fully Homomorphic Encryption (137), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (127), Approximate Homomorphic Encryption with Reduced Approximation Error (123), High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (96), Better Bootstrapping for Approximate Homomorphic Encryption (78), A Full RNS Variant of Approximate Homomorphic Encryption (74), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (71), On the precision loss in approximate homomorphic encryption (65), Cheddar: A Swift Fully Homomorphic Encryption Library Designed for GPU Architectures (57), OpenFHE: Open-Source Fully Homomorphic Encryption Library (52), Revisiting Homomorphic Encryption Schemes for Finite Fields (47), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (47), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (41), Accelerating HE Operations from Key Decomposition Technique (39), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (38), HEonGPU: a GPU-based Fully Homomorphic Encryption Library 1.0 (37), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (37), ModHE: Modular Homomorphic Encryption Using Module Lattices: Potentials and Limitations (36), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (35), Private and reliable neural network inference (31), General Functional Bootstrapping using CKKS (24), Privft: Private and fast text classification with homomorphic encryption (23), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (23), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (20), On the Security of Homomorphic Encryption on Approximate Numbers (18), Secure transformer inference made non-interactive (18), A scalable multi-GPU framework for encrypted large-model inference (18), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (17), F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (16), Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (13), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (13), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (12), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (12), EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computation (12), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (11), SHIP: A Shallow and Highly Parallelizable CKKS Bootstrapping Algorithm (11), THEMIS: A Co-Designed System for Encrypted Transformer Inference (10), CKKS Bootstrapping in less than 10ms (9), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (9), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (9), Security guidelines for implementing homomorphic encryption (8), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (8), BASALISC: Programmable Hardware Accelerator for BGV Fully Homomorphic Encryption (8), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (8), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (8), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (8), Trace-Factored BigSwitch for Matrix-Friendly FHE (7), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (7), Leveraging Discrete CKKS to Bootstrap in High Precision (7), Bootstrapping GBFV with CKKS (6), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (6), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (6), TensorFHE: Achieving Practical Computation on Encrypted Data Using GPGPU (6), FIDESlib: A Fully-Fledged Open-Source FHE Library for Efficient CKKS on GPUs (6), Low-Depth Bootstrapping for Matrix-Native FHE (6), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (6), ENSI: Efficient non-interactive secure inference for large language models (6), Secure outsourced matrix computation and application to neural networks (5), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (5), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (5) …

Polynomial approximation (108 papers)

Efficient Homomorphic Evaluation on Large Intervals (47), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (18), Activate Me!: Designing Efficient Activation Functions for Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (15), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (15), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (15), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (15), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (13), Encryption-friendly LLM architecture (13), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (12), Fast and Accurate Homomorphic Softmax Evaluation (10), Nimbus: Secure and Efficient Two-Party Inference for Transformers (9), Bootstrapping Bits with CKKS (9), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (9), SoK: Private LLM Inference using Approximate Homomorphic Encryption (9), Homomorphic Encryption for Machine Learning and Artificial Intelligence Applications (8), META-BTS: Bootstrapping Precision Beyond the Limit (7), Converting transformers to polynomial form for secure inference over homomorphic encryption (7), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (6), Ring packing and amortized FHEW bootstrapping (6), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (6), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (6), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (6), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (6), Private pathological assessment via machine learning and homomorphic encryption (6), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (6), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (5), Private svm inference on encrypted data (5), Transciphering Framework for Approximate Homomorphic Encryption (5), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (5), Delphi: A Cryptographic Inference Service for Neural Networks (5), Better Bootstrapping for Approximate Homomorphic Encryption (5), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (5), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (5), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (5), SoK: Private transformer-based model inference (4), The Beginner's Textbook for Fully Homomorphic Encryption (4), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (4), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (4), BLEACH: Cleaning Errors in Discrete Computations over CKKS (4), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (4), Demystifying bootstrapping in fully homomorphic encryption (3), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (3), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (3), Faster bootstrapping with polynomial error (3), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (3), PaCo: Bootstrapping for CKKS via Partial CoeffToSlot (3), Leveraging Discrete CKKS to Bootstrap in High Precision (3), General Functional Bootstrapping using CKKS (3), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (3), THEMIS: A Co-Designed System for Encrypted Transformer Inference (3), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (3), ENSI: Efficient non-interactive secure inference for large language models (3), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (3), ARION: Attention-optimized transformer inference on encrypted data (3), THOR: Secure transformer inference with homomorphic encryption (3), Security guidelines for implementing homomorphic encryption (2), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (2), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (2), Privft: Private and fast text classification with homomorphic encryption (2), ModHE: Modular Homomorphic Encryption Using Module Lattices: Potentials and Limitations (2) …

ReLU (102 papers)

Activate Me!: Designing Efficient Activation Functions for Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (80), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (67), Delphi: A Cryptographic Inference Service for Neural Networks (53), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (50), CrypTFlow2: Practical 2-Party Secure Inference (44), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (43), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (37), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (32), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (31), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (28), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (27), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (25), THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryption (25), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (22), Optimized Privacy-Preserving CNN Inference With Fully Homomorphic Encryption (22), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (20), Encryption-friendly LLM architecture (19), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (18), STIP: Efficient and secure non-interactive transformer inference via compact packing (18), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (17), SIGMA: Secure GPT Inference with Function Secret Sharing (17), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (17), Converting transformers to polynomial form for secure inference over homomorphic encryption (16), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (15), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (15), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (13), TFHE Gets Real: An Efficient and Flexible Homomorphic Floating-Point Arithmetic (13), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (12), CrypTen: Secure Multi-Party Computation Meets Machine Learning (12), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (11), Theory, analysis, and best practices for sigmoid selfattention (10), Homomorphic Encryption for Machine Learning and Artificial Intelligence Applications (10), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (10), MPC-Minimized Secure LLM Inference (10), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (9), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (9), MPCFormer: Fast, Performant and Private Transformer Inference with MPC (9), SoK: Privacy-preserving computation techniques for deep learning (8), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (8), Fully Homomorphically Encrypted Deep Learning as a Service (8), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (8), Private and reliable neural network inference (7), Neuracrypt is not private (7), SoK: Private transformer-based model inference (7), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (7), zkLLM: Zero Knowledge Proofs for Large Language Models (7), Programmable Bootstrapping Enables Efficient Homomorphic Inference of Deep Neural Networks (7), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (7), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (6), ABY3: A Mixed Protocol Framework for Machine Learning (6), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (6), Exploring Bitslicing Architectures for Enabling FHE-Assisted Machine Learning (6), An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption (6), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (6), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (5), Iron: Private Inference on Transformers (5), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (5), BLEACH: Cleaning Errors in Discrete Computations over CKKS (5), Fully Homomorphic Encryption with Applications to Privacy-Preserving Machine Learning (4), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (4) …

Semi-honest model (94 papers)

SoK: Private transformer-based model inference (66), ABY3: A Mixed Protocol Framework for Machine Learning (41), Ajax: Fast Threshold Fully Homomorphic Encryption without Noise Flooding (35), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (32), MP-SPDZ: A Versatile Framework for Multi-Party Computation (31), A comprehensive survey on secure outsourced computation and its applications (22), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (20), On-the-Fly Multiparty Computation on the Cloud via Multikey Fully Homomorphic Encryption (15), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (14), Multiparty Computation with Low Communication, Computation and Interaction via Threshold FHE (12), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (12), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (11), Two Round Multiparty Computation via Multi-Key FHE (11), SoK: Privacy-preserving computation techniques for deep learning (10), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (10), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (9), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (8), Private svm inference on encrypted data (8), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (8), Evaluating 2-DNF Formulas on Ciphertexts (8), BumbleBee: Secure Two-party Inference Framework for Large Transformers (8), ARION: Attention-optimized transformer inference on encrypted data (8), Low Communication Threshold FHE from Standard (Module-)LWE (7), Non-interactive Fully Encrypted Machine Learning Protocol for Inference (7), zkLLM: Zero Knowledge Proofs for Large Language Models (7), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (7), MPC-Minimized Secure LLM Inference (7), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (6), CrypTen: Secure Multi-Party Computation Meets Machine Learning (6), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (6), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (5), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (5), CrypTFlow2: Practical 2-Party Secure Inference (5), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (5), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (4), FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System (4), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (4), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (4), SIGMA: Secure GPT Inference with Function Secret Sharing (4), Delphi: A Cryptographic Inference Service for Neural Networks (4), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (4), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (4), Private and reliable neural network inference (3), Fully Homomorphic Encryption for Machine Learning (3), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (3), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (3), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (3), Nimbus: Secure and Efficient Two-Party Inference for Transformers (3), MPCFormer: Fast, Performant and Private Transformer Inference with MPC (3), Secure transformer inference made non-interactive (3), Homomorphic encryption standard (2), Privft: Private and fast text classification with homomorphic encryption (2), Function Secret Sharing (2), Multi-key Fully Homomorphic Encryption with Non-Interactive Setup in the Plain Model (2), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (2), Secure human action recognition by encrypted neural network inference (2), PPFLHE: A privacy-preserving federated learning scheme with homomorphic encryption for healthcare data (2), Federated Learning: An approach with Hybrid Homomorphic Encryption (2), Private pathological assessment via machine learning and homomorphic encryption (2), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (2) …

Softmax (93 papers)

CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (255), Theory, analysis, and best practices for sigmoid selfattention (195), Fast and Accurate Homomorphic Softmax Evaluation (100), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (72), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (72), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (71), THEMIS: A Co-Designed System for Encrypted Transformer Inference (68), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (59), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (54), TRICYCLE: Private transformer inference with tricyclic encodings (53), Encryption-friendly LLM architecture (47), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (46), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (44), SIGMA: Secure GPT Inference with Function Secret Sharing (41), Nimbus: Secure and Efficient Two-Party Inference for Transformers (40), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (39), Pointer sentinel mixture models (38), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (38), ARION: Attention-optimized transformer inference on encrypted data (37), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (36), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (35), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (34), THOR: Secure transformer inference with homomorphic encryption (34), THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryption (31), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (31), SoK: Private LLM Inference using Approximate Homomorphic Encryption (30), SoK: Private transformer-based model inference (29), Iron: Private Inference on Transformers (29), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (29), ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs (26), MPCFormer: Fast, Performant and Private Transformer Inference with MPC (26), STIP: Efficient and secure non-interactive transformer inference via compact packing (26), BumbleBee: Secure Two-party Inference Framework for Large Transformers (25), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (25), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (25), Stealing machine learning models via prediction APIs (23), ENSI: Efficient non-interactive secure inference for large language models (23), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (23), [Skyformer: Remodel selfattention with gaussian kernel and nystr](../papers/skyformer-remodel-selfattention-with-gaussian-kernel-and-nystr.md) (22), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (22), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (21), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (20), zkLLM: Zero Knowledge Proofs for Large Language Models (18), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (18), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (18), PUMA: Secure Inference of LLaMA-7B in Five Minutes (18), Converting transformers to polynomial form for secure inference over homomorphic encryption (18), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (17), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (17), Secure transformer inference made non-interactive (14), Large language model inference acceleration: A comprehensive hardware perspective (13), Trace-Factored BigSwitch for Matrix-Friendly FHE (12), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (11), CipherGPT: Secure Two-Party GPT Inference (11), Recursive deep models for semantic compositionality over a sentiment treebank (9), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (9), Privft: Private and fast text classification with homomorphic encryption (8), On Fully Homomorphic Encryption for Privacy-Preserving Deep Learning (8), CrypTen: Secure Multi-Party Computation Meets Machine Learning (8), A scalable multi-GPU framework for encrypted large-model inference (8) …

Modulus switching (92 papers)

Revisiting Homomorphic Encryption Schemes for Finite Fields (49), The Beginner's Textbook for Fully Homomorphic Encryption (42), Ajax: Fast Threshold Fully Homomorphic Encryption without Noise Flooding (20), OpenFHE: Open-Source Fully Homomorphic Encryption Library (19), (Leveled) Fully Homomorphic Encryption without Bootstrapping (18), A Full RNS Variant of Approximate Homomorphic Encryption (16), SPIRAL: Fast, high-rate single-server PIR via FHE composition (14), Private machine learning classification based on fully homomorphic encryption (14), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (12), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (12), LFHE: Fully Homomorphic Encryption with Bootstrapping Key Size Less than a Megabyte (9), Ring packing and amortized FHEW bootstrapping (9), FINAL: Faster FHE Instantiated with NTRU and LWE (9), F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (8), Fully Homomorphic Encryption for Matrix Arithmetic (8), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (6), General Functional Bootstrapping using CKKS (6), Efficient FHEW Bootstrapping with Small Evaluation Keys, and Applications to Threshold Homomorphic Encryption (5), Bootstrapping GBFV with CKKS (5), Bootstrapping for BGV and BFV Revisited (5), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (5), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (5), BASALISC: Programmable Hardware Accelerator for BGV Fully Homomorphic Encryption (5), FHEW: Bootstrapping Homomorphic Encryption in Less Than a Second (5), Approximate Homomorphic Encryption with Reduced Approximation Error (5), Improved Security for a Ring-Based Fully Homomorphic Encryption Scheme (4), On-the-Fly Multiparty Computation on the Cloud via Multikey Fully Homomorphic Encryption (4), Faster Logical Operations from Discrete CKKS (4), Fully Homomorphic Encryption without Modulus Switching from Classical GapSVP (4), Somewhat Practical Fully Homomorphic Encryption (4), FINALLY: A Multi-Key FHE Scheme Based on NTRU and LWE (3), High-Precision Exact FHE Made Simple, General, and Fast (3), Fully Homomorphic Encryption for Machine Learning (3), Homomorphic Encryption from Learning with Errors: Conceptually-Simpler, Asymptotically-Faster, Attribute-Based (3), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (3), Homomorphic Encryption for Arithmetic of Approximate Numbers (3), Homomorphic encryption standard (2), SoK: Private transformer-based model inference (2), Multiparty Computation with Low Communication, Computation and Interaction via Threshold FHE (2), Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (2), Fully Homomorphic Encryption for Cyclotomic Prime Moduli (2), Fully Homomorphically Encrypted Deep Learning as a Service (2), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (2), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (2), Better Bootstrapping for Approximate Homomorphic Encryption (2), Secure outsourced matrix computation and application to neural networks (1), Secure numerical simulations using fully homomorphic encryption (1), Security guidelines for implementing homomorphic encryption (1), A comprehensive survey on secure outsourced computation and its applications (1), One server for the price of two: Simple and fast single-server private information retrieval (1), Demystifying bootstrapping in fully homomorphic encryption (1), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (1), Pasta: A Case for Hybrid Homomorphic Encryption (1), The HHE Land: Exploring the Landscape of Hybrid Homomorphic Encryption (1), Transciphering Framework for Approximate Homomorphic Encryption (1), Multi-Key Homomorphic Encryption from TFHE (1), Multi-key Fully Homomorphic Encryption with Non-Interactive Setup in the Plain Model (1), Generalized BGV, BFV, and CKKS for Homomorphic Encryption over Matrix Rings (1), Faster bootstrapping with polynomial error (1), Fully Homomorphic Encryption on the Ring of Gaussian Periods (1) …

Non-interactive (91 papers)

On-the-Fly Multiparty Computation on the Cloud via Multikey Fully Homomorphic Encryption (40), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (23), Secure transformer inference made non-interactive (20), Threshold Cryptosystems From Threshold Fully Homomorphic Encryption (15), Fully Homomorphic Encryption with Chosen-Ciphertext Security from LWE (15), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (14), SoK: Private LLM Inference using Approximate Homomorphic Encryption (14), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (12), ARION: Attention-optimized transformer inference on encrypted data (12), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (10), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (10), ENSI: Efficient non-interactive secure inference for large language models (9), TRICYCLE: Private transformer inference with tricyclic encodings (9), STIP: Efficient and secure non-interactive transformer inference via compact packing (8), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (8), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (7), Fully Homomorphic Encryption for Machine Learning (7), THEMIS: A Co-Designed System for Encrypted Transformer Inference (7), SIGMA: Secure GPT Inference with Function Secret Sharing (6), Encryption-friendly LLM architecture (6), Multiparty Computation with Low Communication, Computation and Interaction via Threshold FHE (5), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (5), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (5), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (5), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (5), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (5), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (5), SoK: Private transformer-based model inference (4), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (4), Two Round Multiparty Computation via Multi-Key FHE (4), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (4), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (4), zkLLM: Zero Knowledge Proofs for Large Language Models (4), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (4), CrypTen: Secure Multi-Party Computation Meets Machine Learning (4), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (4), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (4), Benaloh's Dense Probabilistic Encryption Revisited (3), Private and reliable neural network inference (3), A comprehensive survey on secure outsourced computation and its applications (3), Evaluating 2-DNF Formulas on Ciphertexts (3), A Generalisation, a Simplification and Some Applications of Paillier's Probabilistic Public-Key System (3), Ajax: Fast Threshold Fully Homomorphic Encryption without Noise Flooding (3), Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (3), Non-interactive Fully Encrypted Machine Learning Protocol for Inference (3), THOR: Secure transformer inference with homomorphic encryption (3), Secure outsourced matrix computation and application to neural networks (2), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (2), Breaking the Circuit Size Barrier for Secure Computation Under DDH (2), Multi-key Fully Homomorphic Encryption with Non-Interactive Setup in the Plain Model (2), CKKS Bootstrapping in less than 10ms (2), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (2), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (2), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (2), Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering (2), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (2), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (2), BLEACH: Cleaning Errors in Discrete Computations over CKKS (2), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (2), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (2) …

Compiler (88 papers)

EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computation (54), A scalable multi-GPU framework for encrypted large-model inference (42), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (25), Multi-Identity and Multi-Key Leveled FHE from Learning with Errors (25), MP-SPDZ: A Versatile Framework for Multi-Party Computation (25), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (23), F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (22), Homomorphic Encryption from Learning with Errors: Conceptually-Simpler, Asymptotically-Faster, Attribute-Based (19), SoK: Private transformer-based model inference (18), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (18), SoK: Privacy-preserving computation techniques for deep learning (14), Multiparty Computation with Low Communication, Computation and Interaction via Threshold FHE (12), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (12), OpenFHE: Open-Source Fully Homomorphic Encryption Library (11), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (10), Security guidelines for implementing homomorphic encryption (9), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (9), BASALISC: Programmable Hardware Accelerator for BGV Fully Homomorphic Encryption (9), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (8), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (7), SoK: Private LLM Inference using Approximate Homomorphic Encryption (7), On-the-Fly Multiparty Computation on the Cloud via Multikey Fully Homomorphic Encryption (6), Optimized Privacy-Preserving CNN Inference With Fully Homomorphic Encryption (5), Trace-Factored BigSwitch for Matrix-Friendly FHE (5), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (5), Private and reliable neural network inference (4), Breaking the Circuit Size Barrier for Secure Computation Under DDH (4), Building Privacy-Preserving AI system: AI Inference on Encrypted Data with Fully Homomorphic Encryption (4), On the Security of Homomorphic Encryption on Approximate Numbers (4), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (3), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (3), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (3), Secure human action recognition by encrypted neural network inference (3), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (3), Secure transformer inference made non-interactive (3), One server for the price of two: Simple and fast single-server private information retrieval (2), Demystifying bootstrapping in fully homomorphic encryption (2), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (2), Stream ciphers: A Practical Solution for Efficient Homomorphic-Ciphertext Compression (2), Large-Plaintext Functional Bootstrapping in FHE with Small Bootstrapping Keys (2), A Survey on Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (2), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (2), TREBUCHET: Fully Homomorphic Encryption Accelerator for Deep Computation (2), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (2), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (2), CrypTFlow2: Practical 2-Party Secure Inference (2), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (2), Large language model inference acceleration: A comprehensive hardware perspective (1), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (1), Privft: Private and fast text classification with homomorphic encryption (1), Pasta: A Case for Hybrid Homomorphic Encryption (1), Rubato : Noisy Ciphers for Approximate Homomorphic Encryption (Full Version) (1), Transciphering Framework for Approximate Homomorphic Encryption (1), Fully Homomorphic Encryption with Relatively Small Key and Ciphertext Sizes (1), Fully Homomorphic Encryption for Machine Learning (1), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (1), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (1), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (1), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (1), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (1) …

Two-party computation (88 papers)

SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (73), MPC-Minimized Secure LLM Inference (49), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (41), ABY3: A Mixed Protocol Framework for Machine Learning (31), Nimbus: Secure and Efficient Two-Party Inference for Transformers (29), BumbleBee: Secure Two-party Inference Framework for Large Transformers (28), SIGMA: Secure GPT Inference with Function Secret Sharing (27), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (19), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (19), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (19), SoK: Private transformer-based model inference (16), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (16), Breaking the Circuit Size Barrier for Secure Computation Under DDH (15), CipherGPT: Secure Two-Party GPT Inference (15), Delphi: A Cryptographic Inference Service for Neural Networks (13), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (13), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (12), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (11), CrypTFlow2: Practical 2-Party Secure Inference (11), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (10), A comprehensive survey on secure outsourced computation and its applications (9), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (9), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (9), Multiparty Computation with Low Communication, Computation and Interaction via Threshold FHE (8), MP-SPDZ: A Versatile Framework for Multi-Party Computation (8), CrypTen: Secure Multi-Party Computation Meets Machine Learning (8), SoK: Privacy-preserving computation techniques for deep learning (7), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (7), Function Secret Sharing (6), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (6), Secure transformer inference made non-interactive (6), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (5), Multiparty Computation from Somewhat Homomorphic Encryption (5), FheFL: Fully Homomorphic Encryption Friendly Privacy-Preserving Federated Learning with Byzantine Users (5), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (5), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (5), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (4), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (4), THEMIS: A Co-Designed System for Encrypted Transformer Inference (4), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (4), MPCFormer: Fast, Performant and Private Transformer Inference with MPC (3), Iron: Private Inference on Transformers (3), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (3), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (3), Practical secure inference algorithm for a fine-tuned large language model based on fully homomorphic encryption (3), TRICYCLE: Private transformer inference with tricyclic encodings (3), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (3), Secure outsourced matrix computation and application to neural networks (2), Rubato : Noisy Ciphers for Approximate Homomorphic Encryption (Full Version) (2), Two Round Multiparty Computation via Multi-Key FHE (2), Fully Homomorphic Encryption for Machine Learning (2), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (2), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (2), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (2), EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computation (2), InferDPT: Privacy-Preserving Inference for Closed-box Large Language Model (2), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (2), PUMA: Secure Inference of LLaMA-7B in Five Minutes (2), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (2), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (2) …

NTT (75 papers)

TensorFHE: Achieving Practical Computation on Encrypted Data Using GPGPU (118), BASALISC: Programmable Hardware Accelerator for BGV Fully Homomorphic Encryption (93), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (87), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (80), F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (71), Accelerating HE Operations from Key Decomposition Technique (66), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (57), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (40), Cheddar: A Swift Fully Homomorphic Encryption Library Designed for GPU Architectures (37), ModHE: Modular Homomorphic Encryption Using Module Lattices: Potentials and Limitations (32), HEonGPU: a GPU-based Fully Homomorphic Encryption Library 1.0 (30), FIDESlib: A Fully-Fledged Open-Source FHE Library for Efficient CKKS on GPUs (30), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (25), Nimbus: Secure and Efficient Two-Party Inference for Transformers (23), TREBUCHET: Fully Homomorphic Encryption Accelerator for Deep Computation (20), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (20), A scalable multi-GPU framework for encrypted large-model inference (20), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (19), CKKS Bootstrapping in less than 10ms (18), The Beginner's Textbook for Fully Homomorphic Encryption (18), Fully Homomorphic Encryption for Matrix Arithmetic (17), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (15), Approximate Homomorphic Encryption over the Conjugate-invariant Ring (14), Low-Depth Bootstrapping for Matrix-Native FHE (13), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (13), OpenFHE: Open-Source Fully Homomorphic Encryption Library (11), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (11), Revisiting Homomorphic Encryption Schemes for Finite Fields (11), A Full RNS Variant of Approximate Homomorphic Encryption (11), Bootstrapping GBFV with CKKS (8), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (7), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (6), Better Bootstrapping for Approximate Homomorphic Encryption (6), Efficient FHEW Bootstrapping with Small Evaluation Keys, and Applications to Threshold Homomorphic Encryption (5), High-Precision Exact FHE Made Simple, General, and Fast (5), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (5), Efficient Homomorphic Integer Computer from CKKS (5), Trace-Factored BigSwitch for Matrix-Friendly FHE (5), SPIRAL: Fast, high-rate single-server PIR via FHE composition (4), Bootstrapping Bits with CKKS (4), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (3), Secure human action recognition by encrypted neural network inference (3), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (3), FHEW: Bootstrapping Homomorphic Encryption in Less Than a Second (3), Pasta: A Case for Hybrid Homomorphic Encryption (2), Transciphering Framework for Approximate Homomorphic Encryption (2), LFHE: Fully Homomorphic Encryption with Bootstrapping Key Size Less than a Megabyte (2), Faster Logical Operations from Discrete CKKS (2), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (2), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (2), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (2), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (2), Leveraging Discrete CKKS to Bootstrap in High Precision (2), General Functional Bootstrapping using CKKS (2), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (2), Approximate Homomorphic Encryption with Reduced Approximation Error (2), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (2), Secure outsourced matrix computation and application to neural networks (1), On the precision loss in approximate homomorphic encryption (1), Secure numerical simulations using fully homomorphic encryption (1) …

Sigmoid (75 papers)

Theory, analysis, and best practices for sigmoid selfattention (479), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (31), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (26), Private pathological assessment via machine learning and homomorphic encryption (24), Fully Homomorphically Encrypted Deep Learning as a Service (20), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (19), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (16), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (16), ENSI: Efficient non-interactive secure inference for large language models (13), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (12), TFHE Gets Real: An Efficient and Flexible Homomorphic Floating-Point Arithmetic (12), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (11), Fully Homomorphic Encryption for Machine Learning (9), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (9), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (8), Machine Learning Training on Encrypted Data with TFHE (7), Exploring the future of privacy-preserving heart disease prediction: a fully homomorphic encryption-driven logistic regression approach (7), CKKS Bootstrapping in less than 10ms (6), zkLLM: Zero Knowledge Proofs for Large Language Models (6), CrypTen: Secure Multi-Party Computation Meets Machine Learning (6), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (5), Activate Me!: Designing Efficient Activation Functions for Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (5), SIGMA: Secure GPT Inference with Function Secret Sharing (5), A Full RNS Variant of Approximate Homomorphic Encryption (5), SoK: Private LLM Inference using Approximate Homomorphic Encryption (5), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (4), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (4), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (4), Evaluation of Privacy-aware Support Vector Machine (SVM) Learning using Homomorphic Encryption (4), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (4), Iron: Private Inference on Transformers (4), SoK: Privacy-preserving computation techniques for deep learning (3), Private svm inference on encrypted data (3), Towards Case-Optimized Hybrid Homomorphic Encryption (3), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (3), Practical Considerations of Fully Homomorphic Encryption in Privacy-Preserving Machine Learning (3), TREBUCHET: Fully Homomorphic Encryption Accelerator for Deep Computation (3), CipherGPT: Secure Two-Party GPT Inference (3), Secure outsourced matrix computation and application to neural networks (2), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (2), The Beginner's Textbook for Fully Homomorphic Encryption (2), Fully Homomorphic Encryption with Applications to Privacy-Preserving Machine Learning (2), Exploring Bitslicing Architectures for Enabling FHE-Assisted Machine Learning (2), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (2), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (2), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (2), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (2), MP-SPDZ: A Versatile Framework for Multi-Party Computation (2), BumbleBee: Secure Two-party Inference Framework for Large Transformers (2), PUMA: Secure Inference of LLaMA-7B in Five Minutes (2), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (2), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (2), A comprehensive survey on secure outsourced computation and its applications (1), Stealing machine learning models via prediction APIs (1), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (1), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (1), SoK: Private transformer-based model inference (1), A Survey on Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (1), Homomorphic Encryption for Machine Learning and Artificial Intelligence Applications (1), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (1) …

Fine-tuning (74 papers)

Encryption-friendly LLM architecture (69), MPC-Minimized Secure LLM Inference (66), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (61), Practical secure inference algorithm for a fine-tuned large language model based on fully homomorphic encryption (43), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (30), Split-and-Denoise: Protect Large Language Model Inference with Local Differential Privacy (22), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (17), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (16), Integrating large language models in financial investments and market analysis: A survey (14), BERT: Pre-training of deep bidirectional transformers for language understanding (11), PrivateLoRA for Efficient Privacy Preserving LLM (11), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (9), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (9), Cheddar: A Swift Fully Homomorphic Encryption Library Designed for GPU Architectures (8), SoK: Private transformer-based model inference (7), Nimbus: Secure and Efficient Two-Party Inference for Transformers (7), SoK: Private LLM Inference using Approximate Homomorphic Encryption (7), Beyond memorization: Violating privacy via inference with large language models (6), Large language models in medicine (6), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (6), Theory, analysis, and best practices for sigmoid selfattention (5), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (5), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (5), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (5), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (5), Large language model inference acceleration: A comprehensive hardware perspective (4), Improving language understanding by generative pretraining (4), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (4), BumbleBee: Secure Two-party Inference Framework for Large Transformers (4), THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryption (4), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (4), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (4), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (4), TRICYCLE: Private transformer inference with tricyclic encodings (4), ARION: Attention-optimized transformer inference on encrypted data (4), (a)i am not a lawyer, but...: Engaging legal experts towards responsible llm policies for legal advice (3), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (3), zkLLM: Zero Knowledge Proofs for Large Language Models (3), ENSI: Efficient non-interactive secure inference for large language models (3), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (3), GPT-4 System Card (2), Bitnet: Scaling 1-bit transformers for large language models (2), Llama: Open and efficient foundation language models (2), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (2), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (2), Touch of Privacy: A Homomorphic Encryption-Powered Deep Learning Framework for Fingerprint Authentication (2), Private pathological assessment via machine learning and homomorphic encryption (2), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (2), InferDPT: Privacy-Preserving Inference for Closed-box Large Language Model (2), SIGMA: Secure GPT Inference with Function Secret Sharing (2), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (2), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (2), THOR: Secure transformer inference with homomorphic encryption (2), Cost-efficient llm training with lifetimeaware tensor offloading via gpudirect storage (1), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (1), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (1), Private svm inference on encrypted data (1), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (1), Practical Considerations of Fully Homomorphic Encryption in Privacy-Preserving Machine Learning (1), Building Privacy-Preserving AI system: AI Inference on Encrypted Data with Fully Homomorphic Encryption (1) …

Relinearization (74 papers)

Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (111), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (61), ModHE: Modular Homomorphic Encryption Using Module Lattices: Potentials and Limitations (54), Trace-Factored BigSwitch for Matrix-Friendly FHE (50), The Beginner's Textbook for Fully Homomorphic Encryption (49), Somewhat Practical Fully Homomorphic Encryption (34), Generalized BGV, BFV, and CKKS for Homomorphic Encryption over Matrix Rings (32), EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computation (30), HEonGPU: a GPU-based Fully Homomorphic Encryption Library 1.0 (24), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (21), TRICYCLE: Private transformer inference with tricyclic encodings (21), Homomorphic Encryption from Learning with Errors: Conceptually-Simpler, Asymptotically-Faster, Attribute-Based (19), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (19), On-the-Fly Multiparty Computation on the Cloud via Multikey Fully Homomorphic Encryption (18), Private machine learning classification based on fully homomorphic encryption (14), Revisiting Homomorphic Encryption Schemes for Finite Fields (12), High-Precision Arithmetic in Homomorphic Encryption (11), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (8), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (8), Leveraging Discrete CKKS to Bootstrap in High Precision (8), Fully Homomorphic Encryption for Matrix Arithmetic (7), Cryptanalysis of Symmetric Primitives over Rings and a Key Recovery Attack on Rubato (6), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (6), Low-Depth Bootstrapping for Matrix-Native FHE (6), THEMIS: A Co-Designed System for Encrypted Transformer Inference (6), The HHE Land: Exploring the Landscape of Hybrid Homomorphic Encryption (5), Multi-key Fully Homomorphic Encryption with Non-Interactive Setup in the Plain Model (5), Secure numerical simulations using fully homomorphic encryption (4), Fully Homomorphic Encryption on the Ring of Gaussian Periods (4), Stay in Your Lane: Fast Arithmetic over Large Finite Rings with CKKS (4), Phantom: A CUDA-Accelerated Word-Wise Homomorphic Encryption Library (4), Accelerating HE Operations from Key Decomposition Technique (4), Bootstrapping Bits with CKKS (4), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (4), THOR: Secure transformer inference with homomorphic encryption (4), Homomorphic encryption standard (3), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (3), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (3), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (3), TenSEAL: A Library for Encrypted Tensor Operations Using Homomorphic Encryption (3), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (3), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (3), SoK: Private LLM Inference using Approximate Homomorphic Encryption (3), SoK: Privacy-preserving computation techniques for deep learning (2), Privft: Private and fast text classification with homomorphic encryption (2), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (2), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (2), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (2), Evaluation of Privacy-Preserving Support Vector Machine (SVM) Learning Using Homomorphic Encryption (2), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (2), (Leveled) Fully Homomorphic Encryption without Bootstrapping (2), SHIP: A Shallow and Highly Parallelizable CKKS Bootstrapping Algorithm (2), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (2), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (2), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (2), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (2), Security guidelines for implementing homomorphic encryption (1), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (1), Improved Security for a Ring-Based Fully Homomorphic Encryption Scheme (1), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (1) …

Quantization (73 papers)

Large language model inference acceleration: A comprehensive hardware perspective (233), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (148), An Efficient and Privacy-Preserving Federated Learning Approach Based on Homomorphic Encryption (90), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (57), Bitnet: Scaling 1-bit transformers for large language models (53), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (47), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (42), Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering (40), Machine Learning Training on Encrypted Data with TFHE (35), Exploring Bitslicing Architectures for Enabling FHE-Assisted Machine Learning (32), An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption (32), PrivateLoRA for Efficient Privacy Preserving LLM (23), Federated Learning: An approach with Hybrid Homomorphic Encryption (22), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (20), zkLLM: Zero Knowledge Proofs for Large Language Models (20), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (19), Building Privacy-Preserving AI system: AI Inference on Encrypted Data with Fully Homomorphic Encryption (17), SoK: Private transformer-based model inference (15), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (15), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (14), Slalom: Fast, Verifiable and Private Execution of Neural Networks in Trusted Hardware (14), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (10), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (5), Towards Case-Optimized Hybrid Homomorphic Encryption (5), Practical Considerations of Fully Homomorphic Encryption in Privacy-Preserving Machine Learning (5), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (5), Theory, analysis, and best practices for sigmoid selfattention (4), SoK: Privacy-preserving computation techniques for deep learning (4), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (4), Nimbus: Secure and Efficient Two-Party Inference for Transformers (4), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (3), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (3), Federated Learning with Quantum Computing and Fully Homomorphic Encryption: A Novel Computing Paradigm Shift in Privacy-Preserving ML (3), Touch of Privacy: A Homomorphic Encryption-Powered Deep Learning Framework for Fingerprint Authentication (3), MP-SPDZ: A Versatile Framework for Multi-Party Computation (3), CrypTen: Secure Multi-Party Computation Meets Machine Learning (3), Iron: Private Inference on Transformers (3), CrypTFlow2: Practical 2-Party Secure Inference (3), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (3), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (2), FheFL: Fully Homomorphic Encryption Friendly Privacy-Preserving Federated Learning with Byzantine Users (2), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (2), BumbleBee: Secure Two-party Inference Framework for Large Transformers (2), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (2), Trace-Factored BigSwitch for Matrix-Friendly FHE (2), ENSI: Efficient non-interactive secure inference for large language models (2), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (2), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (2), Secure outsourced matrix computation and application to neural networks (1), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (1), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (1), Privft: Private and fast text classification with homomorphic encryption (1), Beyond memorization: Violating privacy via inference with large language models (1), Multi-Key Homomorphic Encryption from TFHE (1), Faster Logical Operations from Discrete CKKS (1), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (1), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (1), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (1), InferDPT: Privacy-Preserving Inference for Closed-box Large Language Model (1), Confidential Computing on NVIDIA Hopper GPUs: A Performance Benchmark Study (1) …

Lookup table (70 papers)

The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (140), General Functional Bootstrapping using CKKS (136), PEGASUS: Bridging Polynomial and Non-polynomial Evaluations in Homomorphic Encryption (78), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (65), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (55), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (50), SIGMA: Secure GPT Inference with Function Secret Sharing (41), TFHE: Fast Fully Homomorphic Encryption over the Torus (40), The Beginner's Textbook for Fully Homomorphic Encryption (34), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (28), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (27), Large language model inference acceleration: A comprehensive hardware perspective (21), CipherGPT: Secure Two-Party GPT Inference (19), TFHE Gets Real: An Efficient and Flexible Homomorphic Floating-Point Arithmetic (19), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (18), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (17), zkLLM: Zero Knowledge Proofs for Large Language Models (15), Towards Case-Optimized Hybrid Homomorphic Encryption (14), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (13), Cryptanalysis of Symmetric Primitives over Rings and a Key Recovery Attack on Rubato (13), The HHE Land: Exploring the Landscape of Hybrid Homomorphic Encryption (10), Faster Logical Operations from Discrete CKKS (10), Large-Plaintext Functional Bootstrapping in FHE with Small Bootstrapping Keys (9), SoK: Private transformer-based model inference (7), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (7), Efficient Homomorphic Integer Computer from CKKS (7), Programmable Bootstrapping Enables Efficient Homomorphic Inference of Deep Neural Networks (5), Demystifying bootstrapping in fully homomorphic encryption (4), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (4), Security guidelines for implementing homomorphic encryption (3), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (3), Multi-Key Homomorphic Encryption from TFHE (3), Machine Learning Training on Encrypted Data with TFHE (3), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (3), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (3), Secure numerical simulations using fully homomorphic encryption (2), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (2), Multi-Identity and Multi-Key Leveled FHE from Learning with Errors (2), ModHE: Modular Homomorphic Encryption Using Module Lattices: Potentials and Limitations (2), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (2), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (2), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (2), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (2), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (2), BLEACH: Cleaning Errors in Discrete Computations over CKKS (2), SPIRAL: Fast, high-rate single-server PIR via FHE composition (1), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (1), Pasta: A Case for Hybrid Homomorphic Encryption (1), LFHE: Fully Homomorphic Encryption with Bootstrapping Key Size Less than a Megabyte (1), Fully Homomorphic Encryption for Cyclotomic Prime Moduli (1), CKKS Bootstrapping in less than 10ms (1), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (1), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (1), Non-interactive Fully Encrypted Machine Learning Protocol for Inference (1), Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering (1), HHEML: Hybrid Homomorphic Encryption for Privacy-Preserving Machine Learning on Edge (1), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (1), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (1), Application of Machine Learning-NLP Approach with Fully Homomorphic Encryption Techniques in Medical PII Data (1), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (1) …

Malicious model (62 papers)

Multiparty Computation with Low Communication, Computation and Interaction via Threshold FHE (48), ABY3: A Mixed Protocol Framework for Machine Learning (37), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (33), On-the-Fly Multiparty Computation on the Cloud via Multikey Fully Homomorphic Encryption (33), Two Round Multiparty Computation via Multi-Key FHE (29), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (24), Ajax: Fast Threshold Fully Homomorphic Encryption without Noise Flooding (15), MP-SPDZ: A Versatile Framework for Multi-Party Computation (15), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (13), SoK: Private transformer-based model inference (10), Non-interactive Fully Encrypted Machine Learning Protocol for Inference (10), A comprehensive survey on secure outsourced computation and its applications (7), SoK: Privacy-preserving computation techniques for deep learning (7), CrypTen: Secure Multi-Party Computation Meets Machine Learning (7), MPC-Minimized Secure LLM Inference (7), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (6), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (6), Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (5), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (5), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (4), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (4), SIGMA: Secure GPT Inference with Function Secret Sharing (4), Fully Homomorphic Encryption for Machine Learning (3), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (3), FheFL: Fully Homomorphic Encryption Friendly Privacy-Preserving Federated Learning with Byzantine Users (3), Federated Learning: An approach with Hybrid Homomorphic Encryption (3), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (3), ARION: Attention-optimized transformer inference on encrypted data (3), Breaking the Circuit Size Barrier for Secure Computation Under DDH (2), Cryptanalysis of Symmetric Primitives over Rings and a Key Recovery Attack on Rubato (2), Revisiting Shamir Secret Sharing for Threshold Fully Homomorphic Encryption (2), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (2), PUMA: Secure Inference of LLaMA-7B in Five Minutes (2), CrypTFlow2: Practical 2-Party Secure Inference (2), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (2), One server for the price of two: Simple and fast single-server private information retrieval (1), Private svm inference on encrypted data (1), Function Secret Sharing (1), Evaluating 2-DNF Formulas on Ciphertexts (1), The HHE Land: Exploring the Landscape of Hybrid Homomorphic Encryption (1), GG-GSW: Chosen-Ciphertext Secure Leveled FHE From Gadget Trapdoors (1), Bootstrapping for BGV and BFV Revisited (1), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (1), Machine Learning Training on Encrypted Data with TFHE (1), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (1), PPFLHE: A privacy-preserving federated learning scheme with homomorphic encryption for healthcare data (1), An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption (1), Private pathological assessment via machine learning and homomorphic encryption (1), EVA: An Encrypted Vector Arithmetic Language and Compiler for Efficient Homomorphic Computation (1), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (1), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (1), Iron: Private Inference on Transformers (1), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (1), BLEACH: Cleaning Errors in Discrete Computations over CKKS (1), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (1), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (1), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (1), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (1), STIP: Efficient and secure non-interactive transformer inference via compact packing (1), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (1) …

LayerNorm (59 papers)

MOAI: Module-optimizing architecture for non-interactive secure transformer inference (53), STIP: Efficient and secure non-interactive transformer inference via compact packing (35), Theory, analysis, and best practices for sigmoid selfattention (34), ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs (32), SIGMA: Secure GPT Inference with Function Secret Sharing (32), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (31), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (31), Iron: Private Inference on Transformers (31), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (27), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (24), Converting transformers to polynomial form for secure inference over homomorphic encryption (24), ARION: Attention-optimized transformer inference on encrypted data (22), SoK: Private LLM Inference using Approximate Homomorphic Encryption (22), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (21), TRICYCLE: Private transformer inference with tricyclic encodings (21), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (20), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (19), THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryption (19), CipherGPT: Secure Two-Party GPT Inference (18), PUMA: Secure Inference of LLaMA-7B in Five Minutes (17), Secure transformer inference made non-interactive (16), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (15), SoK: Private transformer-based model inference (14), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (13), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (11), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (11), THOR: Secure transformer inference with homomorphic encryption (11), THEMIS: A Co-Designed System for Encrypted Transformer Inference (10), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (8), Encryption-friendly LLM architecture (8), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (7), BumbleBee: Secure Two-party Inference Framework for Large Transformers (7), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (7), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (7), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (7), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (6), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (6), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (5), Bitnet: Scaling 1-bit transformers for large language models (4), Attention is all you need (4), zkLLM: Zero Knowledge Proofs for Large Language Models (4), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (4), Nimbus: Secure and Efficient Two-Party Inference for Transformers (4), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (4), Improving language understanding by generative pretraining (2), MPC-Minimized Secure LLM Inference (2), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (2), Practical secure inference algorithm for a fine-tuned large language model based on fully homomorphic encryption (2), Large language model inference acceleration: A comprehensive hardware perspective (1), Llama: Open and efficient foundation language models (1), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (1), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (1), SoK: Privacy-preserving computation techniques for deep learning (1), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (1), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (1), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (1), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (1), Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference (1), ENSI: Efficient non-interactive secure inference for large language models (1)

Attention (58 papers)

Theory, analysis, and best practices for sigmoid selfattention (56), [Skyformer: Remodel selfattention with gaussian kernel and nystr](../papers/skyformer-remodel-selfattention-with-gaussian-kernel-and-nystr.md) (52), Attention is all you need (42), STIP: Efficient and secure non-interactive transformer inference via compact packing (28), zkLLM: Zero Knowledge Proofs for Large Language Models (20), Large language model inference acceleration: A comprehensive hardware perspective (17), Practical secure inference algorithm for a fine-tuned large language model based on fully homomorphic encryption (17), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (15), MPC-Minimized Secure LLM Inference (14), TRICYCLE: Private transformer inference with tricyclic encodings (14), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (12), CipherGPT: Secure Two-Party GPT Inference (11), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (11), Encryption-friendly LLM architecture (11), THOR: Secure transformer inference with homomorphic encryption (11), BERT: Pre-training of deep bidirectional transformers for language understanding (10), SoK: Private transformer-based model inference (10), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (10), ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs (9), Converting transformers to polynomial form for secure inference over homomorphic encryption (9), ENSI: Efficient non-interactive secure inference for large language models (9), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (9), ARION: Attention-optimized transformer inference on encrypted data (9), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (7), Iron: Private Inference on Transformers (6), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (6), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (5), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (5), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (5), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (4), SIGMA: Secure GPT Inference with Function Secret Sharing (4), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (4), SoK: Private LLM Inference using Approximate Homomorphic Encryption (4), Improving language understanding by generative pretraining (3), CrypTen: Secure Multi-Party Computation Meets Machine Learning (3), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (3), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (3), Secure transformer inference made non-interactive (3), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (3), Pointer sentinel mixture models (2), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (2), Split-and-Denoise: Protect Large Language Model Inference with Local Differential Privacy (2), Nimbus: Secure and Efficient Two-Party Inference for Transformers (2), BumbleBee: Secure Two-party Inference Framework for Large Transformers (2), Trace-Factored BigSwitch for Matrix-Friendly FHE (2), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (2), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (2), Bitnet: Scaling 1-bit transformers for large language models (1), fairseq: A fast, extensible toolkit for sequence modeling (1), Llama: Open and efficient foundation language models (1), Privft: Private and fast text classification with homomorphic encryption (1), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (1), Fully Homomorphically Encrypted Deep Learning as a Service (1), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (1), PrivateLoRA for Efficient Privacy Preserving LLM (1), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (1), Fully Homomorphic Encryption for Matrix Arithmetic (1), THEMIS: A Co-Designed System for Encrypted Transformer Inference (1)

GELU (55 papers)

SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (72), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (64), SIGMA: Secure GPT Inference with Function Secret Sharing (60), Nimbus: Secure and Efficient Two-Party Inference for Transformers (53), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (49), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (48), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (43), BumbleBee: Secure Two-party Inference Framework for Large Transformers (43), CipherGPT: Secure Two-Party GPT Inference (43), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (33), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (25), Secure transformer inference made non-interactive (25), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (25), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (24), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (24), Iron: Private Inference on Transformers (24), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (22), PUMA: Secure Inference of LLaMA-7B in Five Minutes (21), MPCFormer: Fast, Performant and Private Transformer Inference with MPC (20), SoK: Private transformer-based model inference (18), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (18), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (18), Converting transformers to polynomial form for secure inference over homomorphic encryption (18), THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryption (17), ARION: Attention-optimized transformer inference on encrypted data (16), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (15), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (15), THOR: Secure transformer inference with homomorphic encryption (15), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (14), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (14), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (13), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (13), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (12), THEMIS: A Co-Designed System for Encrypted Transformer Inference (12), SoK: Private LLM Inference using Approximate Homomorphic Encryption (12), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (11), ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs (10), TRICYCLE: Private transformer inference with tricyclic encodings (10), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (9), STIP: Efficient and secure non-interactive transformer inference via compact packing (9), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (7), CKKS Bootstrapping in less than 10ms (6), zkLLM: Zero Knowledge Proofs for Large Language Models (4), A scalable multi-GPU framework for encrypted large-model inference (4), MPC-Minimized Secure LLM Inference (3), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (3), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (2), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (2), Practical secure inference algorithm for a fine-tuned large language model based on fully homomorphic encryption (2), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (2), Theory, analysis, and best practices for sigmoid selfattention (1), Improving language understanding by generative pretraining (1), BERT: Pre-training of deep bidirectional transformers for language understanding (1), Machine Learning Training on Encrypted Data with TFHE (1), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (1)

Minimax approximation (54 papers)

High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (83), Efficient Homomorphic Evaluation on Large Intervals (71), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (28), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (15), Error Propagation-Aware Scale Design for Efficient Homomorphic Encryption-Based LLM Inference (6), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (6), Fast and Accurate Homomorphic Softmax Evaluation (5), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (4), Improved Bootstrapping for Approximate Homomorphic Encryption (4), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (4), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (3), META-BTS: Bootstrapping Precision Beyond the Limit (3), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (3), Encryption-friendly LLM architecture (3), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (2), Transciphering Framework for Approximate Homomorphic Encryption (2), CKKS Bootstrapping in less than 10ms (2), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (2), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (2), High-Precision Arithmetic in Homomorphic Encryption (2), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (2), BLEACH: Cleaning Errors in Discrete Computations over CKKS (2), Approximate Homomorphic Encryption with Reduced Approximation Error (2), SoK: Private LLM Inference using Approximate Homomorphic Encryption (2), Privft: Private and fast text classification with homomorphic encryption (1), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (1), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (1), Bootstrapping for BGV and BFV Revisited (1), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (1), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (1), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (1), The Beginner's Textbook for Fully Homomorphic Encryption (1), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (1), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (1), Practical Considerations of Fully Homomorphic Encryption in Privacy-Preserving Machine Learning (1), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (1), Optimized Privacy-Preserving CNN Inference With Fully Homomorphic Encryption (1), ARK: Fully Homomorphic Encryption Accelerator with Runtime Data Generation and Inter-Operation Key Reuse (1), BTS: An Accelerator for Bootstrappable Fully Homomorphic Encryption (1), FIDESlib: A Fully-Fledged Open-Source FHE Library for Efficient CKKS on GPUs (1), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (1), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (1), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (1), Accelerating HE Operations from Key Decomposition Technique (1), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (1), Leveraging Discrete CKKS to Bootstrap in High Precision (1), Bootstrapping Bits with CKKS (1), Homomorphic Encryption for Arithmetic of Approximate Numbers (1), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (1), ENSI: Efficient non-interactive secure inference for large language models (1), STIP: Efficient and secure non-interactive transformer inference via compact packing (1), Secure transformer inference made non-interactive (1), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (1), ARION: Attention-optimized transformer inference on encrypted data (1)

Chebyshev (49 papers)

Improved Bootstrapping for Approximate Homomorphic Encryption (42), Better Bootstrapping for Approximate Homomorphic Encryption (27), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (16), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (14), Evaluation of Privacy-aware Support Vector Machine (SVM) Learning using Homomorphic Encryption (12), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (12), Activate Me!: Designing Efficient Activation Functions for Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (11), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (11), High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (10), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (10), THEMIS: A Co-Designed System for Encrypted Transformer Inference (10), ARION: Attention-optimized transformer inference on encrypted data (8), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (7), Private pathological assessment via machine learning and homomorphic encryption (7), Evaluation of Privacy-Preserving Support Vector Machine (SVM) Learning Using Homomorphic Encryption (6), Grafting: Decoupled Scale Factors and Modulus in RNS-CKKS (6), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (6), ENSI: Efficient non-interactive secure inference for large language models (6), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (5), Efficient Homomorphic Evaluation on Large Intervals (5), General Functional Bootstrapping using CKKS (5), SoK: Private LLM Inference using Approximate Homomorphic Encryption (5), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (4), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (4), TRICYCLE: Private transformer inference with tricyclic encodings (4), Transciphering Framework for Approximate Homomorphic Encryption (3), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (3), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (3), Faster bootstrapping with polynomial error (2), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (2), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (2), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (2), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (2), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (2), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (1), SoK: Privacy-preserving computation techniques for deep learning (1), Demystifying bootstrapping in fully homomorphic encryption (1), Stay in Your Lane: Fast Arithmetic over Large Finite Rings with CKKS (1), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (1), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (1), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (1), Over 100x Faster Bootstrapping in Fully Homomorphic Encryption through Memory-centric Optimization with GPUs (1), FIDESlib: A Fully-Fledged Open-Source FHE Library for Efficient CKKS on GPUs (1), CrypTen: Secure Multi-Party Computation Meets Machine Learning (1), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (1), Fast and Accurate Homomorphic Softmax Evaluation (1), META-BTS: Bootstrapping Precision Beyond the Limit (1), Leveraging Discrete CKKS to Bootstrap in High Precision (1), Bootstrapping Bits with CKKS (1)

Federated learning (48 papers)

Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (52), PPFLHE: A privacy-preserving federated learning scheme with homomorphic encryption for healthcare data (48), Secure federated learning applied to medical imaging with fully homomorphic encryption (39), FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System (34), Federated Learning: An approach with Hybrid Homomorphic Encryption (32), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (31), Federated Learning with Quantum Computing and Fully Homomorphic Encryption: A Novel Computing Paradigm Shift in Privacy-Preserving ML (28), An Efficient and Privacy-Preserving Federated Learning Approach Based on Homomorphic Encryption (28), An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption (27), Secure Federated Learning With Fully Homomorphic Encryption for IoT Communications (25), FheFL: Fully Homomorphic Encryption Friendly Privacy-Preserving Federated Learning with Byzantine Users (21), Fully Homomorphic Encryption with Applications to Privacy-Preserving Machine Learning (17), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (14), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (14), Private svm inference on encrypted data (10), THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryption (7), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (6), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (6), SoK: Privacy-preserving computation techniques for deep learning (6), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (5), Exploring the future of privacy-preserving heart disease prediction: a fully homomorphic encryption-driven logistic regression approach (4), PrivateLoRA for Efficient Privacy Preserving LLM (4), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (3), HEonGPU: a GPU-based Fully Homomorphic Encryption Library 1.0 (3), Split-and-Denoise: Protect Large Language Model Inference with Local Differential Privacy (3), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (3), Revisiting Shamir Secret Sharing for Threshold Fully Homomorphic Encryption (2), Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (2), Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering (2), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (2), Private pathological assessment via machine learning and homomorphic encryption (2), Secure numerical simulations using fully homomorphic encryption (1), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (1), SoK: Private transformer-based model inference (1), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (1), Beyond memorization: Violating privacy via inference with large language models (1), The HHE Land: Exploring the Landscape of Hybrid Homomorphic Encryption (1), Machine Learning Training on Encrypted Data with TFHE (1), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (1), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (1), Fully Homomorphically Encrypted Deep Learning as a Service (1), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (1), Evaluation of Privacy-Preserving Support Vector Machine (SVM) Learning Using Homomorphic Encryption (1), Evaluation of Privacy-aware Support Vector Machine (SVM) Learning using Homomorphic Encryption (1), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (1), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (1), Multiparty Homomorphic Encryption from Ring-Learning-with-Errors (1), STIP: Efficient and secure non-interactive transformer inference via compact packing (1)

Taylor expansion (42 papers)

The Beginner's Textbook for Fully Homomorphic Encryption (13), Better Bootstrapping for Approximate Homomorphic Encryption (5), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (5), CHIMERA: Combining Ring-LWE-based Fully Homomorphic Encryption Schemes (4), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (3), CrypTen: Secure Multi-Party Computation Meets Machine Learning (3), Nimbus: Secure and Efficient Two-Party Inference for Transformers (3), BumbleBee: Secure Two-party Inference Framework for Large Transformers (3), Fast and Accurate Homomorphic Softmax Evaluation (3), Bootstrapping for Approximate Homomorphic Encryption (3), General Functional Bootstrapping using CKKS (3), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (2), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (2), PPFLHE: A privacy-preserving federated learning scheme with homomorphic encryption for healthcare data (2), MPCFormer: Fast, Performant and Private Transformer Inference with MPC (2), THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryption (2), Improved Bootstrapping for Approximate Homomorphic Encryption (2), Homomorphic Encryption for Arithmetic of Approximate Numbers (2), Secure numerical simulations using fully homomorphic encryption (1), A comprehensive survey on secure outsourced computation and its applications (1), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (1), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (1), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (1), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (1), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (1), Fully Homomorphically Encrypted Deep Learning as a Service (1), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (1), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (1), PUMA: Secure Inference of LLaMA-7B in Five Minutes (1), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (1), TFHE Gets Real: An Efficient and Flexible Homomorphic Floating-Point Arithmetic (1), PaCo: Bootstrapping for CKKS via Partial CoeffToSlot (1), META-BTS: Bootstrapping Precision Beyond the Limit (1), High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (1), Efficient Bootstrapping for Approximate Homomorphic Encryption with Non-Sparse Keys (1), Bootstrapping Bits with CKKS (1), A Full RNS Variant of Approximate Homomorphic Encryption (1), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (1), STIP: Efficient and secure non-interactive transformer inference via compact packing (1), Secure transformer inference made non-interactive (1), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (1), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (1)

Programmable bootstrapping (33 papers)

TFHE Gets Real: An Efficient and Flexible Homomorphic Floating-Point Arithmetic (70), Towards Case-Optimized Hybrid Homomorphic Encryption (34), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (30), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (24), Programmable Bootstrapping Enables Efficient Homomorphic Inference of Deep Neural Networks (17), Machine Learning Training on Encrypted Data with TFHE (13), High-Precision Exact FHE Made Simple, General, and Fast (11), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (10), Demystifying bootstrapping in fully homomorphic encryption (9), The Beginner's Textbook for Fully Homomorphic Encryption (8), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (7), Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering (7), General Functional Bootstrapping using CKKS (7), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (5), Efficient Homomorphic Integer Computer from CKKS (4), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (4), Plaintext-Ciphertext Matrix Multiplication and FHE Bootstrapping: Fast and Fused (3), The HHE Land: Exploring the Landscape of Hybrid Homomorphic Encryption (2), Efficient Homomorphic Evaluation on Large Intervals (2), Bootstrapping Bits with CKKS (2), Private and reliable neural network inference (1), Secure numerical simulations using fully homomorphic encryption (1), Pasta: A Case for Hybrid Homomorphic Encryption (1), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (1), Large-Plaintext Functional Bootstrapping in FHE with Small Bootstrapping Keys (1), Efficient FHEW Bootstrapping with Small Evaluation Keys, and Applications to Threshold Homomorphic Encryption (1), Bootstrapping GBFV with CKKS (1), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (1), Faster Logical Operations from Discrete CKKS (1), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (1), Exploring Bitslicing Architectures for Enabling FHE-Assisted Machine Learning (1), Optimized Privacy-Preserving CNN Inference With Fully Homomorphic Encryption (1), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (1)

Three-party computation (33 papers)

ABY3: A Mixed Protocol Framework for Machine Learning (58), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (57), MPC-Minimized Secure LLM Inference (20), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (13), SoK: Private transformer-based model inference (7), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (7), CrypTen: Secure Multi-Party Computation Meets Machine Learning (7), MP-SPDZ: A Versatile Framework for Multi-Party Computation (5), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (5), PUMA: Secure Inference of LLaMA-7B in Five Minutes (5), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (4), BumbleBee: Secure Two-party Inference Framework for Large Transformers (4), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (3), CipherGPT: Secure Two-Party GPT Inference (3), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (3), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (2), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (2), CrypTFlow2: Practical 2-Party Secure Inference (2), SoK: Privacy-preserving computation techniques for deep learning (1), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (1), A Survey on Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (1), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (1), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (1), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (1), Nimbus: Secure and Efficient Two-Party Inference for Transformers (1), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (1), SIGMA: Secure GPT Inference with Function Secret Sharing (1), Delphi: A Cryptographic Inference Service for Neural Networks (1), Zephyr: GPU-Efficient Homomorphic Encryption for Privacy-Preserving Transformer Inference (1), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (1), ARION: Attention-optimized transformer inference on encrypted data (1), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (1), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (1)

Functional bootstrapping (26 papers)

General Functional Bootstrapping using CKKS (92), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (40), Large-Plaintext Functional Bootstrapping in FHE with Small Bootstrapping Keys (33), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (20), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (17), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (15), Faster Logical Operations from Discrete CKKS (14), CKKS Bootstrapping in less than 10ms (10), High-Precision Homomorphic ALU over Arbitrary Moduli with O (1) Bootstrapping (9), Efficient Homomorphic Integer Computer from CKKS (5), Stay in Your Lane: Fast Arithmetic over Large Finite Rings with CKKS (4), Fully Homomorphic Encryption for Cyclotomic Prime Moduli (3), Bootstrapping Bits with CKKS (3), Secure numerical simulations using fully homomorphic encryption (2), Security guidelines for implementing homomorphic encryption (2), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (2), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (2), OpenFHE: Open-Source Fully Homomorphic Encryption Library (2), SoK: Private LLM Inference using Approximate Homomorphic Encryption (2), Ajax: Fast Threshold Fully Homomorphic Encryption without Noise Flooding (1), Fully Homomorphic Encryption on the Ring of Gaussian Periods (1), Bootstrapping GBFV with CKKS (1), XBOOT: Free-XOR Gates for CKKS with Applications to Transciphering (1), Efficient Bootstrapping in Fully Homomorphic Encryption for Matrix Arithmetic (1), TFHE Gets Real: An Efficient and Flexible Homomorphic Floating-Point Arithmetic (1), PaCo: Bootstrapping for CKKS via Partial CoeffToSlot (1)

Goldschmidt (24 papers)

HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (21), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (13), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (8), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (8), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (6), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (6), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (4), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (4), THOR: Secure transformer inference with homomorphic encryption (4), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (3), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (3), SoK: Private LLM Inference using Approximate Homomorphic Encryption (3), Secure numerical simulations using fully homomorphic encryption (2), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (2), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (2), MP-SPDZ: A Versatile Framework for Multi-Party Computation (2), Secure transformer inference made non-interactive (2), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (2), ARION: Attention-optimized transformer inference on encrypted data (2), SoK: Private transformer-based model inference (1), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (1), Iron: Private Inference on Transformers (1), THEMIS: A Co-Designed System for Encrypted Transformer Inference (1), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (1)

Remez (24 papers)

High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (92), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (9), Encryption-friendly LLM architecture (9), Converting transformers to polynomial form for secure inference over homomorphic encryption (6), Powerformer: Efficient and high-accuracy privacy-preserving language model with homomorphic encryption (6), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (4), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (4), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (3), CKKS Bootstrapping in less than 10ms (2), Efficient Homomorphic Evaluation on Large Intervals (2), META-BTS: Bootstrapping Precision Beyond the Limit (2), BLEACH: Cleaning Errors in Discrete Computations over CKKS (2), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (2), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (1), Bootstrapping for Approximate Homomorphic Encryption with Negligible Failure-Probability by Using Sparse-Secret Encapsulation (1), The Beginner's Textbook for Fully Homomorphic Encryption (1), DaCapo: Automatic Bootstrapping Management for Efficient Fully Homomorphic Encryption (1), Orion: A Fully Homomorphic Encryption Framework for Deep Learning (1), PrivateLoRA for Efficient Privacy Preserving LLM (1), Fast and Accurate Homomorphic Softmax Evaluation (1), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (1), Leveraging Discrete CKKS to Bootstrap in High Precision (1), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (1), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (1)

Pruning (23 papers)

Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (29), Large language model inference acceleration: A comprehensive hardware perspective (25), An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption (23), MPC-Minimized Secure LLM Inference (21), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (18), Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering (7), Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (5), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (4), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (4), SPIRAL: Fast, high-rate single-server PIR via FHE composition (2), SoK: Privacy-preserving computation techniques for deep learning (2), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (2), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (2), Iron: Private Inference on Transformers (2), Delphi: A Cryptographic Inference Service for Neural Networks (2), ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs (1), [Skyformer: Remodel selfattention with gaussian kernel and nystr](../papers/skyformer-remodel-selfattention-with-gaussian-kernel-and-nystr.md) (1), Rubato : Noisy Ciphers for Approximate Homomorphic Encryption (Full Version) (1), Bootstrapping for BGV and BFV Revisited (1), COMPARISON OF FULLY HOMOMORPHIC ENCRYPTION AND GARBLED CIRCUITS APPROACHES IN PRIVACY-PRESERVING MACHINE LEARNING Kalyan Cheerla Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS July 2025 (1), PrivateLoRA for Efficient Privacy Preserving LLM (1), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (1), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (1)

Encrypted training (22 papers)

Machine Learning Training on Encrypted Data with TFHE (11), Privft: Private and fast text classification with homomorphic encryption (6), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (5), SecretFlow-SPU: A Performant and User-Friendly Framework for Privacy-Preserving Machine Learning (4), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (3), A comprehensive survey on secure outsourced computation and its applications (2), Private svm inference on encrypted data (2), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (2), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (2), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (2), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (2), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (2), On the Security of Homomorphic Encryption on Approximate Numbers (2), A Full RNS Variant of Approximate Homomorphic Encryption (2), Secure outsourced matrix computation and application to neural networks (1), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (1), Practical Considerations of Fully Homomorphic Encryption in Privacy-Preserving Machine Learning (1), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (1), Building Privacy-Preserving AI system: AI Inference on Encrypted Data with Fully Homomorphic Encryption (1), Exploring the future of privacy-preserving heart disease prediction: a fully homomorphic encryption-driven logistic regression approach (1), Programmable Bootstrapping Enables Efficient Homomorphic Inference of Deep Neural Networks (1), Approximate Homomorphic Encryption with Reduced Approximation Error (1)

Paterson-Stockmeyer (20 papers)

Efficient Homomorphic Evaluation on Large Intervals (15), Improved Bootstrapping for Approximate Homomorphic Encryption (8), General Functional Bootstrapping using CKKS (8), The Power of Low Rank: Fast CKKS Functional Bootstrapping for High-Precision Lookup Tables (7), SWIFT: Shallow and SIMD-Aware CKKS Functional Bootstrapping for Low-Latency (5), Better Bootstrapping for Approximate Homomorphic Encryption (5), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (2), Bootstrapping for Approximate Homomorphic Encryption (2), Leveraging Discrete CKKS to Bootstrap in High Precision (2), ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour (2), ENSI: Efficient non-interactive secure inference for large language models (2), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (2), FIDESlib: A Fully-Fledged Open-Source FHE Library for Efficient CKKS on GPUs (1), High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (1), High-Precision Functional Bootstrapping for CKKS from Fourier Extension (1), Homomorphic Encryption for Arithmetic of Approximate Numbers (1), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (1), ARION: Attention-optimized transformer inference on encrypted data (1), THOR: Secure transformer inference with homomorphic encryption (1), SoK: Private LLM Inference using Approximate Homomorphic Encryption (1)

Knowledge distillation (19 papers)

MPCFormer: Fast, Performant and Private Transformer Inference with MPC (6), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (4), CGF-Softmax: A Cumulant-Based Softmax Reformulation for Efficient Inference under Homomorphic Encryption (3), SoK: Private transformer-based model inference (2), InferDPT: Privacy-Preserving Inference for Closed-box Large Language Model (2), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (2), Nimbus: Secure and Efficient Two-Party Inference for Transformers (2), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (2), STIP: Efficient and secure non-interactive transformer inference via compact packing (2), MOAI: Module-optimizing architecture for non-interactive secure transformer inference (2), [Skyformer: Remodel selfattention with gaussian kernel and nystr](../papers/skyformer-remodel-selfattention-with-gaussian-kernel-and-nystr.md) (1), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (1), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (1), Secure federated learning applied to medical imaging with fully homomorphic encryption (1), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (1), MPC-Minimized Secure LLM Inference (1), Ditto: Quantization-aware Secure Inference of Transformers upon MPC (1), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (1), SoK: Private LLM Inference using Approximate Homomorphic Encryption (1)

Membership inference (19 papers)

Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (12), Secure transformer inference made non-interactive (6), Federated Learning: An approach with Hybrid Homomorphic Encryption (3), Exploring the future of privacy-preserving heart disease prediction: a fully homomorphic encryption-driven logistic regression approach (3), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (2), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (2), FheFL: Fully Homomorphic Encryption Friendly Privacy-Preserving Federated Learning with Byzantine Users (2), SoK: Private transformer-based model inference (1), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (1), SoK: Privacy-preserving computation techniques for deep learning (1), ABY3: A Mixed Protocol Framework for Machine Learning (1), An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption (1), Secure federated learning applied to medical imaging with fully homomorphic encryption (1), FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System (1), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (1), An Efficient and Privacy-Preserving Federated Learning Approach Based on Homomorphic Encryption (1), Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference (1), Delphi: A Cryptographic Inference Service for Neural Networks (1), EncFormer: Secure and Efficient Transformer Inference over Encrypted Data (1)

Newton iteration (19 papers)

Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (10), CrypTen: Secure Multi-Party Computation Meets Machine Learning (3), High-Precision Bootstrapping of RNS-CKKS Homomorphic Encryption Using Optimal Minimax Polynomial Approximation and Inverse Sine Function (3), Secure numerical simulations using fully homomorphic encryption (2), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (2), SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC (2), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (2), Converting transformers to polynomial form for secure inference over homomorphic encryption (2), ELLMo: Packing- and Depth-Aware Encrypted Transformer Inference (2), ARION: Attention-optimized transformer inference on encrypted data (2), EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (2), SoK: Private transformer-based model inference (1), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (1), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (1), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (1), BumbleBee: Secure Two-party Inference Framework for Large Transformers (1), Fast and Accurate Homomorphic Softmax Evaluation (1), Encryption-friendly LLM architecture (1), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (1)

Noise flooding (19 papers)

Ajax: Fast Threshold Fully Homomorphic Encryption without Noise Flooding (15), GG-GSW: Chosen-Ciphertext Secure Leveled FHE From Gadget Trapdoors (8), Low Communication Threshold FHE from Standard (Module-)LWE (6), Leveraging Discrete CKKS to Bootstrap in High Precision (5), Security guidelines for implementing homomorphic encryption (2), Collaborative privacy-preserving analysis of oncological data using multiparty homomorphic encryption (1), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (1), Revisiting Shamir Secret Sharing for Threshold Fully Homomorphic Encryption (1), FHE for SIMD Arithmetic Logic Units with Amortized O(1) Bootstrapping per Ciphertext (1), High-Precision Exact FHE Made Simple, General, and Fast (1), Relaxed Functional Bootstrapping: A New Perspective on BGV/BFV Bootstrapping (1), Efficient Homomorphic Integer Computer from CKKS (1), Faster Logical Operations from Discrete CKKS (1), OpenFHE: Open-Source Fully Homomorphic Encryption Library (1), BumbleBee: Secure Two-party Inference Framework for Large Transformers (1), BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers (1), CipherGPT: Secure Two-Party GPT Inference (1), Terrazzo: Memory-Aware GPU Framework for Private Transformer Inference (1), THEMIS: A Co-Designed System for Encrypted Transformer Inference (1)

Square activation (17 papers)

Activate Me!: Designing Efficient Activation Functions for Privacy-Preserving Machine Learning with Fully Homomorphic Encryption (16), Enhancing privacy-preserving machine learning with self-learnable activation functions in fully homomorphic encryption (10), Homomorphic Encryption for Machine Learning in Medicine and Bioinformatics (6), Secure outsourced matrix computation and application to neural networks (5), GAZELLE: A Low Latency Framework for Secure Neural Network Inference (5), Efficient Multi-Key Homomorphic Encryption with Packed Ciphertexts with Application to Oblivious Neural Network Inference (3), Exploring Bitslicing Architectures for Enabling FHE-Assisted Machine Learning (2), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (2), Private and reliable neural network inference (1), Privacy-preserving neural networks with Homomorphic encryption: Challenges and opportunities (1), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (1), Recent advances of privacy-preserving machine learning based on (Fully) Homomorphic Encryption (1), Optimized Privacy-Preserving CNN Inference With Fully Homomorphic Encryption (1), Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data With GPUs (1), TenSEAL: A Library for Encrypted Tensor Operations Using Homomorphic Encryption (1), HeLayers: A Tile Tensors Framework for Large Neural Networks on Encrypted Data (1), Euston: Efficient and User-Friendly Secure Transformer Inference with Non-Interactivity (1)

Secure aggregation (15 papers)

FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System (12), FheFL: Fully Homomorphic Encryption Friendly Privacy-Preserving Federated Learning with Byzantine Users (7), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (5), An Efficient and Privacy-Preserving Federated Learning Approach Based on Homomorphic Encryption (4), Secure federated learning applied to medical imaging with fully homomorphic encryption (3), Low Communication Threshold FHE from Standard (Module-)LWE (2), Revisiting Module Lattice-based Homomorphic Encryption and Application to Secure-MPC (2), Efficiency Optimization Techniques in Privacy-Preserving Federated Learning With Homomorphic Encryption: A Brief Survey (2), Federated Learning: An approach with Hybrid Homomorphic Encryption (2), SoK: Connecting the dots in privacy-preserving ML - systematization of MPC protocols and conversions between secret sharing schemes (1), Revisiting Shamir Secret Sharing for Threshold Fully Homomorphic Encryption (1), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (1), Practical Privacy-Preserving Machine Learning using Fully Homomorphic Encryption (1), PPFLHE: A privacy-preserving federated learning scheme with homomorphic encryption for healthcare data (1), CRAfT: Constant-Round Non-Polynomial Approximation and P2P Network for Secure Transformer Inference (1)

Model inversion (14 papers)

Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (7), Stealing machine learning models via prediction APIs (6), Exploring the future of privacy-preserving heart disease prediction: a fully homomorphic encryption-driven logistic regression approach (6), SoK: Privacy-preserving computation techniques for deep learning (2), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (2), Secure federated learning applied to medical imaging with fully homomorphic encryption (2), FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System (2), BlindFL: Segmented Federated Learning with Fully Homomorphic Encryption (2), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (1), SoK: Private transformer-based model inference (1), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (1), Approximate homomorphic encryption based privacy-preserving machine learning: a survey (1), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (1), A privacy preserving machine learning framework for medical image analysis using quantized fully connected neural networks with TFHE based inference (1)

KV cache (13 papers)

Cachemir: Fully Homomorphic Encrypted Inference of Generative Large Language Model with KV Cache (71), Large language model inference acceleration: A comprehensive hardware perspective (12), Scaling up FHE-based Privacy-Preserving ML: Higher Throughput, Longer Inputs for LLama-3-8B (v1: Scaling up Privacy-Preserving ML: A CKKS Implementation of Llama-2-7B) (10), Towards efficient privacy-preserving machine learning: A systematic review from protocol, model, and system perspectives (9), SoK: Private LLM Inference using Approximate Homomorphic Encryption (7), Privacypreserving llm inference in practice: A comparative survey of techniques, trade-offs, and deployability (6), HEAT: Faster Fully Homomorphic Inference via Approximations-Weights Co-Adaptation (5), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (3), SoK: Private transformer-based model inference (3), Centaur: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference (3), ROSETTA: Efficient and Accurate Privacy-Preserving LLM Decoding via Hybrid CKKS/TFHE Evaluation (3), PrivateLoRA for Efficient Privacy Preserving LLM (1), An Open-Source End-to-End FHE Implementation for Privacy-Preserving Llama 3 8B Inference (1)

Model extraction / stealing (12 papers)

Stealing machine learning models via prediction APIs (34), Practical secure inference algorithm for a fine-tuned large language model based on fully homomorphic encryption (9), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (5), Exploring the future of privacy-preserving heart disease prediction: a fully homomorphic encryption-driven logistic regression approach (4), SIGMA: Secure GPT Inference with Function Secret Sharing (4), Private and reliable neural network inference (3), Privacy-Preserving Machine Learning With Fully Homomorphic Encryption for Deep Neural Network (3), Swift: Fast Secure Neural Network Inference With Fully Homomorphic Encryption (2), Neuracrypt is not private (1), A Survey of Deep Learning Architectures for Privacy-Preserving Machine Learning With Fully Homomorphic Encryption (1), HE-MAN – Homomorphically Encrypted MAchine learning with oNnx models (1), SHAFT: Secure, Handy, Accurate, and Fast Transformer Inference (1)

Bitslicing (8 papers)

Exploring Bitslicing Architectures for Enabling FHE-Assisted Machine Learning (57), DOReN: Toward Efficient Deep Convolutional Neural Networks with Fully Homomorphic Encryption (4), Pasta: A Case for Hybrid Homomorphic Encryption (3), Fully Homomorphic SIMD Operations (3), F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption (3), Large language model inference acceleration: A comprehensive hardware perspective (1), Stream ciphers: A Practical Solution for Efficient Homomorphic-Ciphertext Compression (1), Rasta: A cipher with low ANDdepth and few ANDs per bit (1)

Split learning / split inference (6 papers)

Split-and-Denoise: Protect Large Language Model Inference with Local Differential Privacy (13), SoK: Wildest dreams: Reproducible research in privacy-preserving neural network training (7), PrivateLoRA for Efficient Privacy Preserving LLM (3), Cryptography-based privacy-preserving large language models: a lifecycle survey of frameworks, methods, and future directions (1), Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption (1), InferDPT: Privacy-Preserving Inference for Closed-box Large Language Model (1)