a probabilistic learning algorithm for robust modeling using neural networks with random weights论文.pdfVIP

Information Sciences 313 (2015) 62–78 Contents lists available at ScienceDirect Information Sciences journal homepage: www.else /locate/ins A probabilistic learning algorithm for robust modeling using neural networks with random weights Feilong Cao a, Hailiang Ye a, Dianhui Wang b,c,⇑ a Department of Applied Mathematics, China Jiliang University, Hangzhou 310018, Zhejiang, China b Department of Computer Science and Information Technology, La Trobe University, Melbourne, VIC 3086, Australia c State Key Lab. of Synthetical Automation for Process Industries, Northeastern University, Shenyang 110819, Liaoning, China a r t i c l e i n f o a b s t r a c t Article history: Robust modeling approaches have received considerable attention due to its practical value Received 21 August 2014 to deal with the presence of outliers in data. This paper proposes a probabilistic robust Received in revised form 5 March 2015 learning algorithm for neural networks with random weights (NNRWs) to improve the Accepted 15 March 2015 modeling performance. The robust NNRW model is trained by optimizing a hybrid regular- Available online 1 April 2015 ization loss function according to the sparsity of outliers and compressive sensing theory. The well-known expectation maximization (EM) algorithm is employed to implement our Keywords: proposed algorithm under some assumptions on noise distribution. Experimental results Pro