Privacy-Preserving-Machine-Learning-Resources

Content

• Privacy-Preserving-Machine-Learning-Resources
  • Content
• About
• Secure Machine Learning
• MPC
• Federated Learning
  • Secure Federated Learning
  • Communication Optimization
  • Byzantine-Tolerant
• Privacy Leakages of ML/FL
• Blogs
• Libraries and Frameworks

About

This is a current list of resources related to the research and development of privacy-preserving machine learning.

Secure Machine Learning

• 安全多方计算及其在机器学习中的应用, 计算机研究与发展'21
• Machine Learning Classification over Encrypted Data, NDSS'14
• Oblivious Multi-Party Machine Learning on Trusted Processors, USENIX SECURITY'16
• SecureML: A System for Scalable Privacy-Preserving Machine Learning, S&P'17
• MiniONN: Oblivious Neural Network Predictions via MiniONN Transformations, CCS'17
• Chameleon: A Hybrid Secure Computation Framework for Machine Learning Applications, AsiaCCS'17
• DeepSecure: Scalable Provably-Secure Deep Learning, DAC'17
• Secure Computation for Machine Learning With SPDZ, NIPS'18
• ABY3:a Mixed protocol Framework for Machine Learning, CCS'18
• SecureNN: Efficient and Private Neural Network Training, PoPETs'18
• Gazelle: A Low Latency Framework for Secure Neural Network Inference, USENIX SECURITY'18
• CHET: an optimizing compiler for fully-homomorphic neural-network inferencing, PLDI'19
• New Primitives for Actively-Secure MPC over Rings with Applications to Private Machine Learning, S&P'19
• Helen: Maliciously Secure Coopetitive Learning for Linear Models, S&P'19
• Efficient multi-key homomorphic encryption with packed ciphertexts with application to oblivious neural network inference. CCS'19
• XONN: XNOR-based Oblivious Deep Neural Network Inference, USENIX Security'19
• QUOTIENT: two-party secure neural network training and prediction, CCS'19
• Secure Evaluation of Quantized Neural Networks, PoPETs'20
• ASTRA: High Throughput 3PC over Rings with Application to Secure Prediction, CCSW'19
• SoK: Modular and Efficient Private Decision Tree Evaluation, PoPETs'19
• Trident: Efficient 4PC Framework for Privacy Preserving Machine Learning, NDSS'20
• BLAZE: Blazing Fast Privacy-Preserving Machine Learning, NDSS'20
• FLASH: Fast and Robust Framework for Privacy-preserving Machine Learning, PoPETs'20
• Delphi: A Cryptographic Inference Service for Neural Networks, USENIX SECURITY'20
• ParSecureML: An Efficient Parallel Secure Machine Learning Framework on GPUs, ICPP'20
• FALCON: Honest-Majority Maliciously Secure Framework for Private Deep Learning, PoPETs'21
• MP2ML: A Mixed-Protocol Machine Learning Framework for Private Inference, ARES'20
• SANNS: Scaling Up Secure Approximate k-Nearest Neighbors Search, USENIX Security'20
• PySyft: A Generic Framework for Privacy Preserving Deep Learning
• Private Deep Learning in TensorFlow Using Secure Computation
• CryptoDL: Deep Neural Networks over Encrypted Data
• CryptoNets: Applying Neural Networks to Encrypted Data with High Throughput and Accuracy
• CrypTFlow: Secure TensorFlow Inference
• CrypTFlow2: Practical 2-Party Secure Inference, CCS'20
• ARIANN: Low-Interaction Privacy-Preserving Deep Learning via Function Secret Sharing
• Practical Privacy-Preserving K-means Clustering, PoPETs'20
• SOTERIA: In Search of Efficient Neural Networks for Private Inference, 20
• SWIFT: Super-fast and Robust Privacy-Preserving Machine Learning, USENIX Security'21
• An Efficient 3-Party Framework for Privacy-Preserving Neural Network Inference, ESORICS'20
• Secure and Verifiable Inference in Deep Neural Networks, ACSAC'20
• Privacy-preserving Density-based Clustering, AisaCCS'21
• SIRNN: A Math Library for Secure RNN Inference, S&P'21
• Let’s Stride Blindfolded in a Forest: Sublinear Multi-Client Decision Trees Evaluation, NDSS'21
• MUSE: Secure Inference Resilient to Malicious Clients, USENIX Security'21
• DeepReDuce: ReLU Reduction for Fast Private Inference, ICML'21
• Garbled Neural Networks are Practical
• GForce : GPU-Friendly Oblivious and Rapid Neural Network Inference, USENIX Security'21
• CryptGPU: Fast Privacy-Preserving Machine Learning on the GPU, S&P'21
• GALA : Greedy ComputAtion for Linear Algebra in Privacy-Preserved Neural Networks, NDSS'21
• Fantastic Four: Honest-Majority Four-Party Secure Computation With Malicious Security, USENIX Security'21
• When homomorphic encryption marries secret sharing: secure large-scale sparse logistic regression and applications in risk control, KDD'21
• Glyph: Fast and Accurately Training Deep Neural Networks on Encrypted Data, NeurIPS'20
• Mystique: Efficient Conversions for Zero-Knowledge Proofs with Applications to Machine Learning, USENIX Security'21
• SoK: Efficient Privacy-preserving Clustering, PoPETs'21
• ZEN: Efficient Zero-Knowledge Proofs for Neural Networks
• zkCNN: Zero Knowledge Proofs for Convolutional Neural Network Predictions and Accuracy, CCS'21
• Secure Quantized Training for Deep Learning, ICML
• Cerebro: A Platform for Multi-Party Cryptographic Collaborative Learning, USENIX Security'21
• Tetrad: Actively Secure 4PC for Secure Training and Inference, NDSS'22
• Adam in Private : Secure and Fast Training of Deep Neural Networks with Adaptive Moment Estimation, PoPETs'22
• SIMC: ML Inference Secure Against Malicious Clients at Semi-Honest Cost, USENIX Security'22
• Circa : Stochastic ReLUs for Private Deep Learning, NeurIPS'21
• Banners: Binarized Neural Networks with Replicated Secret Sharing, IH&MMSec'21
• Cheetah: Lean and Fast Secure Two-Party Deep Neural Network Inference, USENIX Security'22
• Secure Poisson Regression, USENIX Security'22
• SecFloat: Accurate Floating-Point meets Secure 2-Party Computation, S&P'22
• MPClan: Protocol Suite for Privacy-Conscious Computations, IACR ePrint'22
• LLAMA: A Low Latency Math Library for Secure Inference, PoPETs'22
• Pika: Secure Computation using Function Secret Sharing over Rings, PoPETs'22
• Piranha: A GPU Platform for Secure Computation, USENIX Security'22
• Efficient Secure Three-Party Sorting with Applications to Data Analysis and Heavy Hitters, CCS'22
• Private and Reliable Neural Network Inference, CCS'22
• SortingHat: Efficient Private Decision Tree Evaluation via Homomorphic Encryption and Transciphering, CCS'22
• Iron: Private Inference on Transformers, NeurIPS '22
• Private and Reliable Neural Network Inference, CCS'22
• SecureBiNN: 3-Party Secure Computation for Binarized Neural Network Inference, ESORICS'22
• [Meteor: Improved Secure 3-Party Neural Network Inference with Reducing Online Communication Costs, https://dl.acm.org/doi/abs/10.1145/3543507.3583272.3583272)
• Private, Efficient, and Accurate: Protecting Models Trained by Multi-party Learning with Differential Privacy, S&P'23
• Bicoptor: Two-round Secure Three-party Non-linear Computation without Preprocessing for Privacy-preserving Machine Learning, S&P'23
• Fusion: Efficient and Secure Inference Resilient to Malicious Servers, NDSS'23
• REDsec: Running Encrypted Discretized Neural Networks in Seconds, NDSS'23
• SoK: Cryptographic Neural-Network Computation, S&P'23
• Secure Floating-Point Training, USENIX Security'23
• Squirrel: A Scalable Secure Two-Party Computation Framework for Training Gradient Boosting Decision Tree, USENIX Security'23
• Efficient 3PC for Binary Circuits with Application to Maliciously-Secure DNN Inference, USENIX Security'23
• Level Up: Private Non-Interactive Decision Tree Evaluation using Levelled Homomorphic Encryption, CCS'23
• PUMA: Secure Inference of LLaMA-7B in Five Minutes
• Orca: FSS-based Secure Training and Inference with GPUs, S&P'24
• SIGMA: Secure GPT Inference with Function Secret Sharing
• CipherGPT: Secure Two-Party GPT Inference
• HELiKs: HE Linear Algebra Kernels for Secure Inference, CCS'23
• MPCDiff: Testing and Repairing MPC-Hardened Deep Learning Models, NDSS'24

MPC

• 实用安全多方计算协议关键技术研究进展, 计算机研究与发展'15
• Scalable and unconditionally secure multiparty computation, Crypto'07
• Sharemind: A framework for fast privacy-preserving computations, ESORICS'08
• Secure computation with fixed-point numbers, FC'10
• Multiparty computation from somewhat homomorphic encryption, Crypto'12
• Practical covertly secure MPC for dishonest majority–or: breaking the SPDZ limits, ESORICS'13
• GMW vs. Yao? Efficient secure two-party computation with low depth circuits, FC'13
• Efficiently Verifiable Computation on Encrypted Data, CCS'14
• ABY: A Framework for Efficient Mixed-Protocol Secure Two-Party Computation, NDSS'15
• MASCOT: faster malicious arithmetic secure computation with oblivious transfer, CCS'16
• High-Throughput Semi-Honest Secure Three-Party Computation with an Honest Majority, CCS'16
• High- throughput secure three-party computation for malicious adversaries and an honest majority, Crypto'17
• A Framework for Constructing Fast MPC over Arithmetic Circuits with Malicious Adversaries and an Honest-Majority, CCS'17
• Overdrive^2k: Making SPDZ Great Again, Eurocrypto'18
• SPDZ^2k: Efficient MPC mod 2^k for Dishonest Majority, Crypto'18
• Secure outsourced matrix computation and application to neural networks, CCS'18
• Fast large-scale honest-majority MPC for malicious adversaries, Crypto'18
• Minimising communication in honest-majority MPC by batchwise multiplication verification, ACNS'18
• PrivPy: General and Scalable Privacy-Preserving Data Mining, KDD'19
• Two-thirds honest-majority MPC for malicious adversaries at almost the cost of semi-honest, CCS'19
• Turbospeedz: Double your online SPDZ! Improving SPDZ using function dependent preprocessing, ACNS'19
• MP-SPDZ: A Versatile Framework for Multi-Party Computation, CCS'20
• Senate: A Maliciously-Secure MPC Platform for Collaborative Analytics, USENIX Security'20
• Improved primitives for mpc over mixed arithmetic-binary circuits, CRYPTO'20
• Malicious Security Comes Free in Honest-Majority MPC, ePrint'20
• MOTION - A Framework for Mixed-Protocol Multi-Party Computation, TOPS'21
• ABY2.0: Improved Mixed-Protocol Secure Two-Party Computation (Full Version), USENIX Security'21
• SynCirc: Efficient Synthesis of Depth-Optimized Circuits for Secure Computation, HOST'21
• MAGE: Nearly Zero-Cost Virtual Memory for Secure Computation, USENIX OSDI'21
• VASA : Vector AES Instructions for Security Applications, ACSAC'21
• ATLAS: Efficient and Scalable MPC in the Honest Majority Setting, CRYPTO'21
• The Cost of IEEE Arithmetic in Secure Computation, LatinCrypt'21
• Fast Fully Secure Multi-Party Computation over Any Ring with Two-Thirds Honest Majority, CCS'22
• NFGen: Automatic Non-linear Function Evaluation Code Generator for General-purpose MPC Platforms, CCS'22
• PentaGOD: Stepping beyond Traditional GOD with Five Parties, CCS'22
• TurboPack: Honest Majority MPC with Constant Online Communication, CCS'22
• Selective MPC: Distributed Computation of Differentially Private Key-Value Statistics, CCS'22
• To Trust or Not to Trust: Hybrid Multi-party Computation with Trusted Execution Environment, NDSS'22
• Binary Search in Secure Computation, NDSS'22
• More Efficient Dishonest Majority Secure Computation over $\mathbb{Z}_{2^k}$ via Galois Rings, Crypto'22
• Sharing Transformation and Dishonest Majority MPC with Packed Secret Sharing, Crypto'22
• Le Mans: Dynamic and Fluid MPC for Dishonest Majority, Crypto'22
• Round-Optimal and Communication-Efficient Multiparty Computation, EUROCRYPT'22
• Practical Non-interactive Publicly Verifiable Secret Sharing with Thousands of Parties, EUROCRYPT'22
• Highly Efficient OT-Based Multiplication Protocols, EUROCRYPT'22
• Batch-OT with Optimal Rate, EUROCRYPT'22
• Round-Optimal Multi-Party Computation with Identifiable Abort, EUROCRYPT'22
• Secure Multiparty Computation with Free Branching, EUROCRYPT'22
• Secure Multiparty Computation with Sublinear Preprocessing, EUROCRYPT'22
• Attaining GOD Beyond Honest Majority With Friends and Foes, Asiacrypt'22
• Polymath: Low-Latency MPC via Secure Polynomial Evaluations and Its Applications, PoPETs'22
• Silph: A Framework for Scalable and Accurate Generation of Hybrid MPC Protocols, S&P'23
• Faster Secure Comparisons with Offline Phase for Efficient Private Set Intersection, NDSS'23
• FLUTE: Fast and Secure Lookup Table Evaluations, S&P'23
• Linear Communication in Malicious Majority MPC, CCS'23
• Grotto: Screaming fast (2+1)-PC or $\mathbb{Z}_{2^n}$ via (2,2)-DPFs, CCS'23
• COMBINE: COMpilation and Backend-INdependent vEctorization for Multi-Party Computation, CCS'23
• Don’t Eject the Impostor: Fast Three-Party Computation With a Known Cheater, IEEE S&P'24
• Scalable Mixed-Mode MPC, IEEE S&P'24

Federated Learning

Secure Federated Learning

• Privacy-Preserving Deep Learning, CCS'15
• Prio: Private, Robust, and Scalable Computation of Aggregate Statistics, NSDI'17
• Practical Secure Aggregation for Privacy Preserving Machine Learning, CCS'17
• Privacy-Preserving Deep Learning via Additively Homomorphic Encryption, TIFS'17
• NIKE-based Fast Privacy-preserving High-dimensional Data Aggregation for Mobile Devices, CACR'18
• PrivFL: Practical Privacy-preserving Federated Regressions on High-dimensional Data over Mobile Networks, CCSW'19
• VerifyNet: Secure and verifiable federated learning, TIFS'19
• PrivColl: Practical Privacy-Preserving Collaborative Machine Learning
• NPMML: A Framework for Non-interactive Privacy-preserving Multi-party Machine Learning, TDSC'20
• SAFER: Sparse secure Aggregation for FEderated leaRning
• Secure Byzantine-Robust Machine Learning
• Secure Single-Server Aggregation with (Poly)Logarithmic Overhead, CCS'20
• Batchcrypt: Efficient homomorphic encryption for cross-silo federated learning, USENIX ATC'21
• FedSel: Federated SGD under Local Differential Privacy with Top-k Dimension Selection, DASFAA'20
• FLGUARD: Secure and Private Federated Learning, Cryptology Eprint'21
• Biscotti: A Blockchain System for Private and Secure Federated Learning, TPDS'21
• POSEIDON: Privacy-Preserving Federated Neural Network Learning, NDSS'21
• PPFL: Privacy-preserving Federated Learning with Trusted Execution Environments, MobiSys'21
• EIFFeL: Ensuring Integrity for Federated Learning, CCS'22
• Efficient Differentially Private Secure Aggregation for Federated Learning via Hardness of Learning with Errors, USENIX Security'22
• Prio+: Privacy Preserving Aggregate Statistics via Boolean Shares, SCN'22
• Local and Central Differential Privacy for Robustness and Privacy in Federated Learning, NDSS'22
• ELSA: Secure Aggregation for Federated Learning with Malicious Actors, S&P'23
• Flamingo: Multi-Round Single-Server Secure Aggregation with Applications to Private Federated Learning, SP'23
• RoFL: Robustness of Secure Federated Learning, SP'23

Communication Optimization

• Terngrad: Ternary gradients to reduce communication in distributed deep learning, NIPS'17
• The Convergence of Sparsified Gradient Methods, NIPS'18

Byzantine-Tolerant

• Machine Learning with Adversaries: Byzantine Tolerant Gradient Descent, NIPS'17
• Byzantine stochastic gradient descent, NIPS'18
• The Hidden Vulnerability of Distributed Learning in Byzantium, ICML'18
• Byzantine-Robust Distributed Learning: Towards Optimal Statistical Rates, ICML'18
• Local Model Poisoning Attacks to Byzantine-Robust Federated Learning, USENIX Security'20
• FLTrust: Byzantine-robust Federated Learning via Trust Bootstrapping, NDSS'21
• Manipulating the Byzantine: Optimizing Model Poisoning Attacks and Defenses for Federated Learning, NDSS'21
• Justinian’s GAAvernor: Robust Distributed Learning with Gradient Aggregation Agent, USENIX Security'20
• Byzantine-robust and privacy-preserving framework for FEDML, ICLR Workshop'21
• Learning from History for Byzantine Robust Optimization, ICML'21
• FLAME: Taming backdoors in federated learning, USENIX Security'22
• BayBFed: Bayesian Backdoor Defense for Federated Learning, SP'23

Privacy Leakages of ML/FL

• Membership inference attacks against machine learning models, S&P'17
• Comprehensive privacy analysis of deep learning: Passive and active white-box inference attacks against centralized and federated learning, S&P'19
• Data Poisoning Attacks Against Federated Learning Systems, ESORICS'20
• A Framework for Evaluating Client Privacy Leakages in Federated Learning, ESORICS'20
• A Critical Overview of Privacy in Machine Learning, IEEE Security & Privacy'21
• Enhanced Membership Inference Attacks against Machine Learning Models, CCS'22

Blogs

• Cryptography and Machine Learning: Mixing both for private data analysis
• Building Safe A.I.: A Tutorial for Encrypted Deep Learning
• Awesome MPC: Curated List of resources for MPC
• 机器学习隐私保护

Libraries and Frameworks

• TinyGarble: Logic Synthesis and Sequential Descriptions for Yao's Garbled Circuits
• SPDZ-2: Multiparty computation with SPDZ, MASCOT, and Overdrive offline phases
• ABY: A Framework for Efficient Mixed-Protocol Secure Two-Party Computation
• Obliv - C: C compiler for embedding privacy preserving protocols:
• TFHE: Fast Fully Homomorphic Encryption Library over the Torus
• SEAL: Simple Encypted Arithmatic Library
• PySEAL: Python interface to SEAL
• HElib: An Implementation of homomorphic encryption
• EzPC: programmable, efficient, and scalable secure two-party computation for machine learning
• CUDA-accelerated Fully Homomorphic Encryption Library
• CrypTen: A framework for Privacy Preserving Machine Learning
• tf-encrypted: A Framework for Machine Learning on Encrypted Data
• Sharemind
• PythonPaillier
• TenSEAL
• MP-SPDZ
• Securenn-public
• SecMML
• mnist-mpc
• Private-Set-Intersection
• falcon-public
• Rosetta
• Antchain-MPC
• Kunlun
• MOTION2NX
• SecureQ8
• mpc-benchmarks
• muse
• Primihub
• concrete
• SecretFlow