Differentially Private Federated Bayesian Optimization with Distributed Exploration

Bayesian optimization (BO) has recently been extended to the federated learning (FL) setting by the federated Thompson sampling (FTS) algorithm, which has promising applications such as federated hyperparameter tuning. However, FTS is not equipped with a rigorous privacy guarantee which is an important consideration in FL. Recent works have incorporated differential privacy (DP) into the training of deep neural networks through a general framework for adding DP to iterative algorithms. Following this general DP framework, our work here integrates DP into FTS to preserve user-level privacy. We also leverage the ability of this general DP framework to handle different parameter vectors, as well as the technique of local modeling for BO, to further improve the utility of our algorithm through distributed exploration (DE). The resulting differentially private FTS with DE (DP-FTS-DE) algorithm is endowed with theoretical guarantees for both the privacy and utility and is amenable to interesting theoretical insights about the privacy-utility trade-off. We also use real-world experiments to show that DP-FTS-DE achieves high utility (competitive performance) with a strong privacy guarantee (small privacy loss) and induces a trade-off between privacy and utility.

[1]  Xiang Li,et al.  On the Convergence of FedAvg on Non-IID Data , 2019, ICLR.

[2]  Nicolas Vayatis,et al.  Parallel Gaussian Process Optimization with Upper Confidence Bound and Pure Exploration , 2013, ECML/PKDD.

[3]  Jinshuo Dong,et al.  Deep Learning with Gaussian Differential Privacy , 2020, Harvard data science review.

[4]  Cong Shen,et al.  Federated Multi-armed Bandits with Personalization , 2021, AISTATS.

[5]  Kian Hsiang Low,et al.  Distributed Batch Gaussian Process Optimization , 2017, ICML.

[6]  Kian Hsiang Low,et al.  Fault-Tolerant Federated Reinforcement Learning with Theoretical Guarantee , 2021, ArXiv.

[7]  Patrick Jaillet,et al.  Top-k Ranking Bayesian Optimization , 2020, AAAI.

[8]  K. H. Low,et al.  Optimizing Conditional Value-At-Risk of Black-Box Functions , 2021, NeurIPS.

[9]  Úlfar Erlingsson,et al.  Prochlo: Strong Privacy for Analytics in the Crowd , 2017, SOSP.

[10]  Yu-Xiang Wang,et al.  Subsampled Rényi Differential Privacy and Analytical Moments Accountant , 2018, AISTATS.

[11]  Patrick Jaillet,et al.  Collaborative Bayesian Optimization with Fair Regret , 2021, ICML.

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

[13]  Jun Zhao,et al.  Local Differential Privacy-Based Federated Learning for Internet of Things , 2020, IEEE Internet of Things Journal.

[14]  Roman Garnett,et al.  Differentially Private Bayesian Optimization , 2015, ICML.

[15]  Kian Hsiang Low,et al.  Decentralized High-Dimensional Bayesian Optimization with Factor Graphs , 2017, AAAI.

[16]  Zhaowei Zhu,et al.  Federated Bandit: A Gossiping Approach , 2021, SIGMETRICS.

[17]  Kian Hsiang Low,et al.  Nonmyopic Gaussian Process Optimization with Macro-Actions , 2020, AISTATS.

[18]  Alán Aspuru-Guzik,et al.  Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Space , 2017, ICML.

[19]  Kian Hsiang Low,et al.  Federated Bayesian Optimization via Thompson Sampling , 2020, NeurIPS.

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

[21]  Bingsheng He,et al.  Model-Contrastive Federated Learning , 2021, 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[22]  Benjamin Recht,et al.  Random Features for Large-Scale Kernel Machines , 2007, NIPS.

[23]  Charles Elkan,et al.  Differential Privacy and Machine Learning: a Survey and Review , 2014, ArXiv.

[24]  Elaine Shi,et al.  Privacy-Preserving Aggregation of Time-Series Data , 2011, NDSS.

[25]  Andreas Krause,et al.  Information-Theoretic Regret Bounds for Gaussian Process Optimization in the Bandit Setting , 2009, IEEE Transactions on Information Theory.

[26]  Bingsheng He,et al.  Practical Federated Gradient Boosting Decision Trees , 2019, AAAI.

[27]  H. Vincent Poor,et al.  Federated Learning With Differential Privacy: Algorithms and Performance Analysis , 2019, IEEE Transactions on Information Forensics and Security.

[28]  Tshilidzi Marwala,et al.  Healing Products of Gaussian Process Experts , 2020, ICML.

[29]  Kian Hsiang Low,et al.  Private Outsourced Bayesian Optimization , 2020, ICML.

[30]  Kian Hsiang Low,et al.  Bayesian Optimization Meets Bayesian Optimal Stopping , 2019, ICML.

[31]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[32]  Ameet Talwalkar,et al.  Federated Multi-Task Learning , 2017, NIPS.

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

[34]  Kirthevasan Kandasamy,et al.  Parallelised Bayesian Optimisation via Thompson Sampling , 2018, AISTATS.

[35]  Moni Naor,et al.  Our Data, Ourselves: Privacy Via Distributed Noise Generation , 2006, EUROCRYPT.

[36]  Adam D. Smith,et al.  Distributed Differential Privacy via Shuffling , 2018, IACR Cryptol. ePrint Arch..

[37]  Gregory Cohen,et al.  EMNIST: Extending MNIST to handwritten letters , 2017, 2017 International Joint Conference on Neural Networks (IJCNN).

[38]  H. Brendan McMahan,et al.  A General Approach to Adding Differential Privacy to Iterative Training Procedures , 2018, ArXiv.

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

[40]  Harold Soh,et al.  Efficient Exploration of Reward Functions in Inverse Reinforcement Learning via Bayesian Optimization , 2020, NeurIPS.

[41]  Kian Hsiang Low,et al.  R2-B2: Recursive Reasoning-Based Bayesian Optimization for No-Regret Learning in Games , 2020, ICML.

[42]  Aaron Roth,et al.  Gaussian differential privacy , 2019, Journal of the Royal Statistical Society: Series B (Statistical Methodology).

[43]  Bingsheng He,et al.  Federated Learning on Non-IID Data Silos: An Experimental Study , 2021, 2022 IEEE 38th International Conference on Data Engineering (ICDE).

[44]  Abhimanyu Dubey,et al.  Differentially-Private Federated Linear Bandits , 2020, NeurIPS.

[45]  Rui Zhang,et al.  A Hybrid Approach to Privacy-Preserving Federated Learning , 2018, Informatik Spektrum.

[46]  S L Warner,et al.  Randomized response: a survey technique for eliminating evasive answer bias. , 1965, Journal of the American Statistical Association.

[47]  Matthias Poloczek,et al.  Scalable Global Optimization via Local Bayesian Optimization , 2019, NeurIPS.

[48]  Bingsheng He,et al.  A Survey on Federated Learning Systems: Vision, Hype and Reality for Data Privacy and Protection , 2019, IEEE Transactions on Knowledge and Data Engineering.

[49]  Richard Nock,et al.  Advances and Open Problems in Federated Learning , 2021, Found. Trends Mach. Learn..

[50]  Svetha Venkatesh,et al.  A Privacy Preserving Bayesian Optimization with High Efficiency , 2018, PAKDD.

[51]  Davide Anguita,et al.  A Public Domain Dataset for Human Activity Recognition using Smartphones , 2013, ESANN.

[52]  Anit Kumar Sahu,et al.  Federated Optimization in Heterogeneous Networks , 2018, MLSys.

[53]  Andreas Krause,et al.  No-Regret Learning in Unknown Games with Correlated Payoffs , 2019, NeurIPS.

[54]  Ruben Martinez-Cantin,et al.  Fully Distributed Bayesian Optimization with Stochastic Policies , 2019, IJCAI.

[55]  Andreas Krause,et al.  Parallelizing Exploration-Exploitation Tradeoffs with Gaussian Process Bandit Optimization , 2012, ICML.

[56]  H. Brendan McMahan,et al.  Learning Differentially Private Recurrent Language Models , 2017, ICLR.

[57]  Aditya Gopalan,et al.  On Kernelized Multi-armed Bandits , 2017, ICML.

[58]  Bingsheng He,et al.  Privacy-Preserving Gradient Boosting Decision Trees , 2019, AAAI.

[59]  Lawrence Carin,et al.  Multi-Task Learning for Classification with Dirichlet Process Priors , 2007, J. Mach. Learn. Res..

[60]  Kirthevasan Kandasamy,et al.  Gaussian Process Bandit Optimisation with Multi-fidelity Evaluations , 2016, NIPS.

[61]  Kian Hsiang Low,et al.  Value-at-Risk Optimization with Gaussian Processes , 2021, ICML.

[62]  Kian Hsiang Low,et al.  Trusted-Maximizers Entropy Search for Efficient Bayesian Optimization , 2021, UAI.

[63]  Sofya Raskhodnikova,et al.  What Can We Learn Privately? , 2008, 2008 49th Annual IEEE Symposium on Foundations of Computer Science.

[64]  Kian Hsiang Low,et al.  Bayesian Optimization with Binary Auxiliary Information , 2019, UAI.

[65]  Jian Tan,et al.  Local Differential Privacy for Bayesian Optimization , 2020, AAAI.

[66]  Glenn Fung,et al.  Predicting Readmission Risk with Institution Specific Prediction Models , 2013, 2013 IEEE International Conference on Healthcare Informatics.

[67]  Daniel Schall,et al.  Evaluation of Hyperparameter-Optimization Approaches in an Industrial Federated Learning System , 2021, ArXiv.

[68]  Bryan Kian Hsiang Low,et al.  Information-Based Multi-Fidelity Bayesian Optimization , 2017 .

[69]  Maria-Florina Balcan,et al.  Federated Hyperparameter Tuning: Challenges, Baselines, and Connections to Weight-Sharing , 2021, NeurIPS.

[70]  Anit Kumar Sahu,et al.  Federated Learning: Challenges, Methods, and Future Directions , 2019, IEEE Signal Processing Magazine.

[71]  Rui Hu,et al.  Personalized Federated Learning With Differential Privacy , 2020, IEEE Internet of Things Journal.

[72]  Nando de Freitas,et al.  Taking the Human Out of the Loop: A Review of Bayesian Optimization , 2016, Proceedings of the IEEE.