Privacy-Preserving-Machine-Learning

This is a paper reading list for privacy-perserving machine learning (PPML), including but not limited to related research on multi-party computation (MPC), Homomorphic Encryption (HE). The list covers related papers, conferences, tools and other resources. These papers are mainly published in 2023, and some classical papers are also collected.

Literatures in this page are arranged by categories and ordered by the time posted online, including the following topics:

Multi-party Computation (MPC)

Homomorphic Encryption (HE)

• [arXiv 202306] Homomorphic Encryption: An Analysis of its Applications in Searchable Encryption [paper]
• [arXiv 202306] NTT-Based Polynomial Modular Multiplication for Homomorphic Encryption: A Tutorial [paper]
• [arXiv 202306] High-Resolution Convolutional Neural Networks on Homomorphically Encrypted Data via Sharding Ciphertexts [paper]
• [arXiv 202302] HyPHEN: A Hybrid Packing Method and Optimizations for Homomorphic Encryption-Based Neural Networks [paper]
• [arXiv 202302] TT-TFHE: a Torus Fully Homomorphic Encryption-Friendly Neural Network Architecture [paper]
• [HPCA 2021] Cheetah: Optimizing and Accelerating Homomorphic Encryption for Private Inference [paper]

Private Inference with Efficient Activations

• [arXiv 202306] PASNet: Polynomial Architecture Search Framework for Two-party Computation-based Secure Neural Network Deployment [paper] (the same with RRNet)
• [arXiv 202306] Fast and Private Inference of Deep Neural Networks by Co-designing Activation Functions [paper]
• [arXiv 202304] DeepReShape: Redesigning Neural Networks for Efficient Private Inference [paper]
• [arXiv 202302] RRNet: Towards ReLU-Reduced Neural Network for Two-party Computation Based Private Inference [paper] (the same with PASNet)
• [arXiv 202301] Learning to Linearize Deep Neural Networks for Secure and Efficient Private Inference [paper]

Private Inference for Transformers/LLM

• [arXiv 202306] LLMs Can Understand Encrypted Prompt: Towards Privacy-Computing Friendly Transformers [paper]
• [arXiv 202306] MERGE: Fast Private Text Generation [paper]
• [arXiv 202305] Differentially Private Attention Computation [paper]
• [arXiv 202303] Primer: Fast Private Transformer Inference on Encrypted Data [paper]

Private Inference with Quantization

• [arXiv 202305] Approximate Private Inference in Quantized Models [paper]

Hardware Accelerator for Private Inference

• [arXiv 202302] RRNet: Towards ReLU-Reduced Neural Network for Two-party Computation Based Private Inference [paper]
• [HPCA 2021] Cheetah: Optimizing and Accelerating Homomorphic Encryption for Private Inference [paper]
• [USENIX S&P 2018] GAZELLE: A Low Latency Framework for Secure Neural Network Inference [paper]