A Privacy-Preserving and Trustable Multi-agent Learning Framework

Distributed multi-agent learning enables agents to cooperatively train a model without requiring to share their datasets. While this setting ensures some level of privacy, it has been shown that, even when data is not directly shared, the training process is vulnerable to privacy attacks including data reconstruction and model inversion attacks. Additionally, malicious agents that train on inverted labels or random data, may arbitrarily weaken the accuracy of the global model. This paper addresses these challenges and presents Privacy-preserving and trustable Distributed Learning (PT-DL), a fully decentralized framework that relies on Differential Privacy to guarantee strong privacy protection of the agents data, and Ethereum smart contracts to ensure trustability. The paper shows that PT-DL is resilient up to a 50% collusion attack, with high probability, in a malicious trust model and the experimental evaluation illustrates the benefits of the proposed model as a privacy-preserving and trustable distributed multi-agent learning system on several classification tasks.

[1]  Pascal Van Hentenryck,et al.  Privacy-Preserving Federated Data Sharing , 2019, AAMAS.

[2]  Daniel S. Kermany,et al.  Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning , 2018, Cell.

[3]  Jakub Konecný,et al.  Federated Optimization: Distributed Optimization Beyond the Datacenter , 2015, ArXiv.

[4]  Mariagrazia Dotoli,et al.  Distributed Alternating Direction Method of Multipliers for Linearly Constrained Optimization Over a Network , 2020, IEEE Control Systems Letters.

[5]  Tianbao Yang,et al.  Trading Computation for Communication: Distributed Stochastic Dual Coordinate Ascent , 2013, NIPS.

[6]  Feng Yan,et al.  LEASGD: an Efficient and Privacy-Preserving Decentralized Algorithm for Distributed Learning , 2018, ArXiv.

[7]  Feng Yan,et al.  Towards Decentralized Deep Learning with Differential Privacy , 2019, CLOUD.

[8]  Tassilo Klein,et al.  Differentially Private Federated Learning: A Client Level Perspective , 2017, ArXiv.

[9]  Yuchen Zhang,et al.  Communication-Efficient Distributed Optimization of Self-Concordant Empirical Loss , 2015, ArXiv.

[10]  Michael I. Jordan,et al.  Adding vs. Averaging in Distributed Primal-Dual Optimization , 2015, ICML.

[11]  Michael Moeller,et al.  Inverting Gradients - How easy is it to break privacy in federated learning? , 2020, NeurIPS.

[12]  Martin J. Wainwright,et al.  Information-theoretic lower bounds for distributed statistical estimation with communication constraints , 2013, NIPS.

[13]  Yue Zhang,et al.  DeepChain: Auditable and Privacy-Preserving Deep Learning with Blockchain-Based Incentive , 2019, IEEE Transactions on Dependable and Secure Computing.

[14]  Peter F. Edemekong,et al.  Health Insurance Portability and Accountability Act , 2020 .

[15]  Ilya Mironov,et al.  Rényi Differential Privacy , 2017, 2017 IEEE 30th Computer Security Foundations Symposium (CSF).

[16]  Enrico Pontelli,et al.  Multi-Variable Agents Decomposition for DCOPs , 2016, AAAI Conference on Artificial Intelligence.

[17]  Blaise Agüera y Arcas,et al.  Communication-Efficient Learning of Deep Networks from Decentralized Data , 2016, AISTATS.

[18]  A. Meyer.  The Health Insurance Portability and Accountability Act. , 1997, Tennessee medicine : journal of the Tennessee Medical Association.

[19]  Joseph Paul Cohen,et al.  COVID-19 Image Data Collection: Prospective Predictions Are the Future , 2020, ArXiv.

[20]  Maria-Florina Balcan,et al.  Distributed Learning, Communication Complexity and Privacy , 2012, COLT.

[21]  Mamun Bin Ibne Reaz,et al.  Can AI Help in Screening Viral and COVID-19 Pneumonia? , 2020, IEEE Access.

[22]  Cynthia Dwork,et al.  Calibrating Noise to Sensitivity in Private Data Analysis , 2006, TCC.

[23]  Pramod Viswanath,et al.  The Composition Theorem for Differential Privacy , 2013, IEEE Transactions on Information Theory.

[24]  Ian Goodfellow,et al.  Deep Learning with Differential Privacy , 2016, CCS.

[25]  Mingyan Liu,et al.  Improving the Privacy and Accuracy of ADMM-Based Distributed Algorithms , 2018, ICML.

[26]  Rachid Guerraoui,et al.  Fast and Differentially Private Algorithms for Decentralized Collaborative Machine Learning , 2017, ArXiv.

[27]  Vitaly Shmatikov,et al.  Membership Inference Attacks Against Machine Learning Models , 2016, 2017 IEEE Symposium on Security and Privacy (SP).

[28]  Daniel Davis Wood,et al.  ETHEREUM: A SECURE DECENTRALISED GENERALISED TRANSACTION LEDGER , 2014 .

[29]  Ohad Shamir,et al.  Distributed stochastic optimization and learning , 2014, 2014 52nd Annual Allerton Conference on Communication, Control, and Computing (Allerton).

[30]  Joseph Paul Cohen,et al.  COVID-19 Image Data Collection , 2020, ArXiv.

[31]  Li Zhang,et al.  Rényi Differential Privacy of the Sampled Gaussian Mechanism , 2019, ArXiv.

[32]  Mehdi Bennis,et al.  On-Device Federated Learning via Blockchain and its Latency Analysis , 2018, ArXiv.

[33]  Jinyuan Jia,et al.  Local Model Poisoning Attacks to Byzantine-Robust Federated Learning , 2019, USENIX Security Symposium.

[34]  Rachid Guerraoui,et al.  Personalized and Private Peer-to-Peer Machine Learning , 2017, AISTATS.

[35]  Aaron Roth,et al.  The Algorithmic Foundations of Differential Privacy , 2014, Found. Trends Theor. Comput. Sci..

[36]  Peter Richtárik,et al.  Fast distributed coordinate descent for non-strongly convex losses , 2014, 2014 IEEE International Workshop on Machine Learning for Signal Processing (MLSP).

[37]  Enrico Pontelli,et al.  Distributed Constraint Optimization Problems and Applications: A Survey , 2016, J. Artif. Intell. Res..

[38]  Somesh Jha,et al.  Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures , 2015, CCS.

[39]  Quanyan Zhu,et al.  Dynamic Differential Privacy for ADMM-Based Distributed Classification Learning , 2017, IEEE Transactions on Information Forensics and Security.