Distributed learning with infinitely many hypotheses

We consider a distributed learning setup where a network of agents sequentially access realizations of a set of random variables with unknown distributions. The network objective is to find a parametrized distribution that best describes their joint observations in the sense of the Kullback-Leibler divergence. We analyze the case of countably many hypotheses and the case of a continuum of hypotheses. We provide non-asymptotic bounds for the concentration rate of the agents' beliefs around the correct hypothesis in terms of the number of agents, the network parameters, and the learning abilities of the agents. Additionally, we provide a novel motivation for a general set of distributed non-Bayesian update rules as instances of the distributed stochastic mirror descent algorithm.

[1]  Michael Athans,et al.  Convergence and asymptotic agreement in distributed decision problems , 1982, 1982 21st IEEE Conference on Decision and Control.

[2]  Changzhi Wu,et al.  Stochastic mirror descent method for distributed multi-agent optimization , 2016, Optimization Letters.

[3]  John N. Tsitsiklis,et al.  On distributed averaging algorithms and quantization effects , 2007, 2008 47th IEEE Conference on Decision and Control.

[4]  Qipeng Liu,et al.  Distributed detection via Bayesian updates and consensus , 2014, 2015 34th Chinese Control Conference (CCC).

[5]  Gerhard P. Hancke,et al.  Opportunities and Challenges of Wireless Sensor Networks in Smart Grid , 2010, IEEE Transactions on Industrial Electronics.

[6]  Douglas Gale,et al.  Bayesian learning in social networks , 2003, Games Econ. Behav..

[7]  Angelia Nedić,et al.  Fast Convergence Rates for Distributed Non-Bayesian Learning , 2015, IEEE Transactions on Automatic Control.

[8]  Pooya Molavi,et al.  Foundations of Non-Bayesian Social Learning , 2016 .

[9]  Alex Olshevsky,et al.  Linear Time Average Consensus on Fixed Graphs and Implications for Decentralized Optimization and Multi-Agent Control , 2014, 1411.4186.

[10]  Marc Teboulle,et al.  Mirror descent and nonlinear projected subgradient methods for convex optimization , 2003, Oper. Res. Lett..

[11]  James A. Bucklew,et al.  Robust decentralized source localization via averaging , 2005, Proceedings. (ICASSP '05). IEEE International Conference on Acoustics, Speech, and Signal Processing, 2005..

[12]  L. Birge,et al.  Model selection via testing: an alternative to (penalized) maximum likelihood estimators , 2006 .

[13]  Qipeng Liu,et al.  Non-Bayesian learning in social networks with time-varying weights , 2011, Proceedings of the 30th Chinese Control Conference.

[14]  Lucien Birg'e.  About the non-asymptotic behaviour of Bayes estimators , 2014 .

[15]  C. A. Rogers.  A note on coverings , 1957 .

[16]  Shahin Shahrampour,et al.  Switching to learn , 2015, 2015 American Control Conference (ACC).

[17]  Shahin Shahrampour,et al.  Exponentially fast parameter estimation in networks using distributed dual averaging , 2013, 52nd IEEE Conference on Decision and Control.

[18]  Daron Acemoglu,et al.  Bayesian Learning in Social Networks , 2008 .

[19]  Nitin H. Vaidya,et al.  Asynchronous Distributed Hypothesis Testing in the Presence of Crash Failures , 2016, ArXiv.

[20]  A. Juditsky,et al.  Learning by mirror averaging , 2005, math/0511468.

[21]  Elchanan Mossel,et al.  Asymptotic learning on Bayesian social networks , 2012, Probability Theory and Related Fields.

[22]  Asuman E. Ozdaglar,et al.  Convergence of rule-of-thumb learning rules in social networks , 2008, 2008 47th IEEE Conference on Decision and Control.

[23]  Angelia Nedic,et al.  Network independent rates in distributed learning , 2015, 2016 American Control Conference (ACC).

[24]  Shahin Shahrampour,et al.  Distributed Detection: Finite-Time Analysis and Impact of Network Topology , 2014, IEEE Transactions on Automatic Control.

[25]  Ilya Dumer.  Covering Spheres with Spheres , 2007, Discret. Comput. Geom..

[26]  Michael G. Rabbat,et al.  Multi-agent mirror descent for decentralized stochastic optimization , 2015, 2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP).

[27]  John Anderson,et al.  Wireless sensor networks for habitat monitoring , 2002, WSNA '02.

[28]  Ali Jadbabaie,et al.  Non-Bayesian Social Learning , 2011, Games Econ. Behav..

[29]  Ali Jadbabaie,et al.  Learning without Recall from actions of neighbors , 2016, 2016 American Control Conference (ACC).

[30]  Robert D. Nowak,et al.  Decentralized source localization and tracking [wireless sensor networks] , 2004, 2004 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[31]  Angelia Nedic,et al.  Nonasymptotic convergence rates for cooperative learning over time-varying directed graphs , 2014, 2015 American Control Conference (ACC).

[32]  Angelia Nedic,et al.  Distributed Optimization Over Time-Varying Directed Graphs , 2015, IEEE Trans. Autom. Control..

[33]  Shahin Shahrampour,et al.  Learning without recall by random walks on directed graphs , 2015, 2015 54th IEEE Conference on Decision and Control (CDC).

[34]  T. Javidi,et al.  Social learning and distributed hypothesis testing , 2014, 2014 IEEE International Symposium on Information Theory.

[35]  Jie Lin,et al.  Coordination of groups of mobile autonomous agents using nearest neighbor rules , 2003, IEEE Trans. Autom. Control..

[36]  Chee-Yee Chong,et al.  Sensor networks: evolution, opportunities, and challenges , 2003, Proc. IEEE.