Asynchronous Decentralized Stochastic Optimization in Heterogeneous Networks

We consider expected risk minimization in multi-agent systems comprised of distinct subsets of agents operating without a common time-scale. Each individual in the network is charged with minimizing the global objective function, which is an average of sum of the statistical average loss function of each agent in the network. Since agents are not assumed to observe data from identical distributions, the hypothesis that all agents seek a common action is violated, and thus the hypothesis upon which consensus constraints are formulated is violated. Thus, we consider nonlinear network proximity constraints which incentivize nearby nodes to make decisions which are close to one another but not necessarily coincide. Moreover, agents are not assumed to receive their sequentially arriving observations on a common time index, and thus seek to learn in an asynchronous manner. An asynchronous stochastic variant of the Arrow-Hurwicz saddle point method is proposed to solve this problem which operates by alternating primal stochastic descent steps and Lagrange multiplier updates which penalize the discrepancies between agents. This tool leads to an implementation that allows for each agent to operate asynchronously with local information only and message passing with neighbors. Our main result establishes that the proposed method yields convergence in expectation both in terms of the primal sub-optimality and constraint violation to radii of sizes $\mathcal{O}(\sqrt{T})$ and $\mathcal{O}(T^{3/4})$, respectively. Empirical evaluation on an asynchronously operating wireless network that manages user channel interference through an adaptive communications pricing mechanism demonstrates that our theoretical results translates well to practice.

[1]  Ketan Rajawat,et al.  Asynchronous Incremental Stochastic Dual Descent Algorithm for Network Resource Allocation , 2017, IEEE Transactions on Signal Processing.

[2]  Alejandro Ribeiro,et al.  Online learning for characterizing unknown environments in ground robotic vehicle models , 2016, 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[3]  Brian M. Sadler,et al.  Proximity Without Consensus in Online Multiagent Optimization , 2016, IEEE Transactions on Signal Processing.

[4]  Cédric Archambeau,et al.  Adaptive Algorithms for Online Convex Optimization with Long-term Constraints , 2015, ICML.

[5]  Alejandro Ribeiro,et al.  D4L: Decentralized dynamic discriminative dictionary learning , 2015, 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[6]  Ali H. Sayed,et al.  Asynchronous adaptive networks , 2015, ArXiv.

[7]  Halim Yanikomeroglu,et al.  Interference-Aware Energy-Efficient Resource Allocation for OFDMA-Based Heterogeneous Networks With Incomplete Channel State Information , 2015, IEEE Transactions on Vehicular Technology.

[8]  Konstantinos I. Tsianos,et al.  Asynchronous decentralized optimization in heterogeneous systems , 2014, 53rd IEEE Conference on Decision and Control.

[9]  Ali H. Sayed,et al.  Stability and Performance Limits of Adaptive Primal-Dual Networks , 2014, IEEE Transactions on Signal Processing.

[10]  Alejandro Ribeiro,et al.  A Saddle Point Algorithm for Networked Online Convex Optimization , 2014, IEEE Transactions on Signal Processing.

[11]  Jie Chen,et al.  Multitask Diffusion Adaptation Over Networks , 2013, IEEE Transactions on Signal Processing.

[12]  Qing Ling,et al.  On the Convergence of Decentralized Gradient Descent , 2013, SIAM J. Optim..

[13]  Alexander Shapiro,et al.  Stochastic Approximation approach to Stochastic Programming , 2013 .

[14]  Alejandro Ribeiro,et al.  Decentralized dynamic optimization through the alternating direction method of multipliers , 2013, 2013 IEEE 14th Workshop on Signal Processing Advances in Wireless Communications (SPAWC).

[15]  Francis R. Bach,et al.  Adaptivity of averaged stochastic gradient descent to local strong convexity for logistic regression , 2013, J. Mach. Learn. Res..

[16]  Asuman E. Ozdaglar,et al.  Distributed Alternating Direction Method of Multipliers , 2012, 2012 IEEE 51st IEEE Conference on Decision and Control (CDC).

[17]  Michael G. Rabbat,et al.  Distributed dual averaging for convex optimization under communication delays , 2012, 2012 American Control Conference (ACC).

[18]  Shai Shalev-Shwartz,et al.  Online Learning and Online Convex Optimization , 2012, Found. Trends Mach. Learn..

[19]  José M. F. Moura,et al.  Fast Distributed Gradient Methods , 2011, IEEE Transactions on Automatic Control.

[20]  Rong Jin,et al.  Trading regret for efficiency: online convex optimization with long term constraints , 2011, J. Mach. Learn. Res..

[21]  Stephen J. Wright,et al.  Hogwild: A Lock-Free Approach to Parallelizing Stochastic Gradient Descent , 2011, NIPS.

[22]  Angelia Nedic,et al.  Distributed Asynchronous Constrained Stochastic Optimization , 2011, IEEE Journal of Selected Topics in Signal Processing.

[23]  Jie Zhang,et al.  OFDMA femtocells: A roadmap on interference avoidance , 2009, IEEE Communications Magazine.

[24]  Angelia Nedic,et al.  Subgradient Methods for Saddle-Point Problems , 2009, J. Optimization Theory and Applications.

[25]  Asuman E. Ozdaglar,et al.  Distributed Subgradient Methods for Multi-Agent Optimization , 2009, IEEE Transactions on Automatic Control.

[26]  Angelia Nedic,et al.  Distributed Stochastic Subgradient Projection Algorithms for Convex Optimization , 2008, J. Optim. Theory Appl..

[27]  Léon Bottou,et al.  The Tradeoffs of Large Scale Learning , 2007, NIPS.

[28]  Daniel Pérez Palomar,et al.  A tutorial on decomposition methods for network utility maximization , 2006, IEEE Journal on Selected Areas in Communications.

[29]  Mung Chiang,et al.  Cross-Layer Congestion Control, Routing and Scheduling Design in Ad Hoc Wireless Networks , 2006, Proceedings IEEE INFOCOM 2006. 25TH IEEE International Conference on Computer Communications.

[30]  M.G. Rabbat,et al.  Generalized consensus computation in networked systems with erasure links , 2005, IEEE 6th Workshop on Signal Processing Advances in Wireless Communications, 2005..

[31]  Simon Haykin,et al.  Cognitive radio: brain-empowered wireless communications , 2005, IEEE Journal on Selected Areas in Communications.

[32]  Martin Zinkevich,et al.  Online Convex Programming and Generalized Infinitesimal Gradient Ascent , 2003, ICML.

[33]  Rich Caruana,et al.  Multitask Learning , 1997, Machine Learning.

[34]  Pierre Priouret,et al.  Adaptive Algorithms and Stochastic Approximations , 1990, Applications of Mathematics.

[35]  John N. Tsitsiklis,et al.  Distributed Asynchronous Deterministic and Stochastic Gradient Optimization Algorithms , 1984, 1984 American Control Conference.

[36]  B. V. Dean,et al.  Studies in Linear and Non-Linear Programming. , 1959 .

[37]  H. Robbins A Stochastic Approximation Method , 1951 .

[38]  Alejandro Ribeiro,et al.  D-MAP: Distributed Maximum a Posteriori Probability Estimation of Dynamic Systems , 2013, IEEE Transactions on Signal Processing.

[39]  Dimitri P. Bertsekas,et al.  Convex Analysis and Optimization , 2003 .

[40]  Vivek S. Borkar,et al.  Distributed Asynchronous Incremental Subgradient Methods , 2001 .

[41]  Vladimir N. Vapnik,et al.  The Nature of Statistical Learning Theory , 2000, Statistics for Engineering and Information Science.

[42]  John N. Tsitsiklis,et al.  Partial Solutions Manual Parallel and Distributed Computation : Numerical Methods , 1997 .