分布式字典学习方法-模式识别与智能系统专业论文.docx

Generated by CamScanner Generated by CamScanner Title: DISTRIBUTED DICTIONARY LEARNING METHOD Major: Pattern Recognition and Intelligent Systems Name: Miaohua ZHANG Supervisor: Prof. Junli LlANG Abstract This dissertation develops a distributed dictionary leaming algorithm for sparse representation of the data distributed across nodes of sensor networks，where the sensitive or private datn are stored or there is no fusion center or there exists a Big Data application. The main c，ontributions of this paper are: 1) We decouple the combined dictionary atom update and nonzero coetlicient revision procedure into two-stage operations to facilitate distributed computations，tirstly updating the dictionary atom in terms of the eigenvalue decomposition of the sum of the residual (correlation) matrices across the nodes then implementing a local projection operation to obtain the related representation coefficients for each node; 2) We cast the aforementioned atom update problem as a set of decentralized optimization subproblems with consensus constraints. Then ，we simplify the Multiplier update for the symmetry undirected graphs in sensor networks and minimize the separable subproblems to attain the consistent estimates iteratively; and 3) Dictionary atoms are typically constrained to be of unit norl11 in order to avoid the scaling ambiguity. We efficiently solve the resultant hidden convex subploblems by determining the optimal Lagrange Multiplier. Some experiments are given to show th3t the proposed algorithm is an alternative distributed dictionary leaming approach ，and is sllitable for thc sensor network environment. Key words: Distributed dictionary learning; Sparse representation; Distributed data; K-SVD; Altcrnating-direction method of multipliers (ADMM); Lagrange Multiplier; Sensor networks; EigcnvalllC Decomposition (EVD); Fusion center; Big Data. This research is Suppolted by National Natural Science Foundation of China(No. 611721