{"responseHeader":{"status":0,"QTime":5,"params":{"q":"{!q.op=AND}id:\"102830\"","hl":"true","hl.simple.post":"","hl.fragsize":"5000","fq":"!embargo_tdt:[NOW TO *]","hl.fl":"ocr_t","hl.method":"unified","wt":"json","hl.simple.pre":""}},"response":{"numFound":1,"start":0,"docs":[{"file_name_t":"Soller-Automated_Detection.pdf","thumb_s":"/d7/0a/d70abb549f7594cd2629bb4589139d2f9e069e9c.jpg","oldid_t":"compsci 10943","setname_s":"ir_computersa","restricted_i":0,"format_t":"application/pdf","modified_tdt":"2016-05-26T00:00:00Z","file_s":"/69/27/69276c1662b1ced3aa72d603262f84694541d4bc.pdf","title_t":"Page 73","ocr_t":"g' (h l,j ) = dgd(hh z,.j ) = g ( hz,j )( 1- g ( hz,j )) . l,J One example of a sigmoid function is 1 g(hz ·) - ---' 3 - 1 + exp ( -hz,j)' 58 (10.9) Funahashi proved that any continuous mapping can be approximated through a multilayer perceptron with one hidden layer whose threshold functions are sigmoid [49]; the multilayer perceptron is a universal approximator. This dissertation focused on classification, which required the artificial neural network to approximate decision regions. Hornik extended Funahashi 's proofs to any continuous, bounded, and non constant threshold function [7 4] . The most common learning algorithm for linearly dependent data patterns and multiple layer perceptron architectures with nonlinear threshold functions is backward error propagation. This algorithm is based on Widrow's Least Mean Squares {LMS) algorithm [183] . It finds a locally optimal solution in the weight space for the following performance measure p J(w) = L Jp(w), (10.10) p=l where Jp( w) is the total squared error for the pth pattern, (10.11) The backward error propagation algorithm implements iterative weight changes in the direction of the negative gradient, aJ(w) P aJp(w) Wz,j,i(k + 1) = W[,j,i(k)- T] a lw(k) = W[,j,i(k)- T] L a lw(k)· W[,j,i p=l W[,j,i (10.12) This is written in terms of the weight change","id":102830,"created_tdt":"2016-05-26T00:00:00Z","parent_i":102961,"_version_":1642982670129430530}]},"highlighting":{"102830":{"ocr_t":[]}}}