Privacy-Preserving Machine Learning: Methods, Challenges and Directions

Abstract

Machine learning (ML) is increasingly being adopted in a wide variety of application domains. Usually, a well-performing ML model relies on a large volume of training data and high-powered computational resources. Such a need for and the use of huge volumes of data raise serious privacy concerns because of the potential risks of leakage of highly privacy-sensitive information; further, the evolving regulatory environments that increasingly restrict access to and use of privacy-sensitive data add significant challenges to fully benefiting from the power of ML for data-driven applications. A trained ML model may also be vulnerable to adversarial attacks such as membership, attribute, or property inference attacks and model inversion attacks. Hence, well-designed privacy-preserving ML (PPML) solutions are critically needed for many emerging applications. Increasingly, significant research efforts from both academia and industry can be seen in PPML areas that aim toward integrating privacy-preserving techniques into ML pipeline or specific algorithms, or designing various PPML architectures. In particular, existing PPML research cross-cut ML, systems and applications design, as well as security and privacy areas; hence, there is a critical need to understand state-of-the-art research, related challenges and a research roadmap for future research in PPML area. In this paper, we systematically review and summarize existing privacy-preserving approaches and propose a Phase, Guarantee, and Utility (PGU) triad based model to understand and guide the evaluation of various PPML solutions by decomposing their privacy-preserving functionalities. We discuss the unique characteristics and challenges of PPML and outline possible research directions that leverage as well as benefit multiple research communities such as ML, distributed systems, security and privacy.

References in corpus (47)

• Distilling the Knowledge in a Neural Network
• Deep Learning with Differential Privacy
• Communication-Efficient Learning of Deep Networks from Decentralized Data
• Language Models are Few-Shot Learners
• Federated Learning: Strategies for Improving Communication Efficiency
• Comprehensive Privacy Analysis of Deep Learning: Passive and Active White-box Inference Attacks against Centralized and Federated Learning
• A Survey on Federated Learning Systems: Vision, Hype and Reality for Data Privacy and Protection
• Renyi Differential Privacy
• BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain
• How To Backdoor Federated Learning
• Differentially Private Federated Learning: A Client Level Perspective
• Distributed Federated Learning for Ultra-Reliable Low-Latency Vehicular Communications
• Private federated learning on vertically partitioned data via entity resolution and additively homomorphic encryption
• iDLG: Improved Deep Leakage from Gradients
• Differentially Private Generative Adversarial Network
• HybridAlpha: An Efficient Approach for Privacy-Preserving Federated Learning
• CryptoDL: Deep Neural Networks over Encrypted Data
• LEAF: A Benchmark for Federated Settings
• Model compression via distillation and quantization
• Threats to Federated Learning: A Survey
• Inverting Gradients -- How easy is it to break privacy in federated learning?
• SecureBoost: A Lossless Federated Learning Framework
• FastSecAgg: Scalable Secure Aggregation for Privacy-Preserving Federated Learning
• Auditing Differentially Private Machine Learning: How Private is Private SGD?
• Federated Learning for Healthcare Informatics
• DataLens: Scalable Privacy Preserving Training via Gradient Compression and Aggregation
• Privacy for Free: Communication-Efficient Learning with Differential Privacy Using Sketches
• PPFL: Privacy-preserving Federated Learning with Trusted Execution Environments
• Data Poisoning against Differentially-Private Learners: Attacks and Defenses
• Enhancing the Privacy of Federated Learning with Sketching
• The Tradeoff Between Privacy and Accuracy in Anomaly Detection Using Federated XGBoost
• Glyph: Fast and Accurately Training Deep Neural Networks on Encrypted Data
• Differentially Private Meta-Learning
• Conditionals in Homomorphic Encryption and Machine Learning Applications
• Assessing differentially private deep learning with Membership Inference
• CryptGPU: Fast Privacy-Preserving Machine Learning on the GPU
• Byzantine-Robust and Privacy-Preserving Framework for FedML
• Adaptive Histogram-Based Gradient Boosted Trees for Federated Learning
• FedSKETCH: Communication-Efficient and Private Federated Learning via Sketching
• Privacy-preserving Learning via Deep Net Pruning
• Mitigating Leakage in Federated Learning with Trusted Hardware
• PPCA: Privacy-preserving Principal Component Analysis Using Secure Multiparty Computation(MPC)
• Citadel: Protecting Data Privacy and Model Confidentiality for Collaborative Learning with SGX
• Separation of Powers in Federated Learning
• Revisiting Model-Agnostic Private Learning: Faster Rates and Active Learning
• SIRNN: A Math Library for Secure RNN Inference
• Revisiting Secure Computation Using Functional Encryption: Opportunities and Research Directions

Cited by in corpus (5)

• A Federated Learning-based Industrial Health Prognostics for Heterogeneous Edge Devices using Matched Feature Extraction
• GME: GPU-based Microarchitectural Extensions to Accelerate Homomorphic Encryption
• Federated Motor Imagery Classification for Privacy-Preserving Brain-Computer Interfaces
• SoK: Systematic analysis of adversarial threats against deep learning approaches for autonomous anomaly detection systems in SDN-IoT networks
• Complex QA and language models hybrid architectures, Survey