霍普菲尔-坦克神经网路之最适化应用【旅行推销员问题】

232HOPFIELDTANKNEURALNETWORKTRAVELLINGSALESMANPROBLEM……ABSTRACTTheartificialneuralnetworkisapartofA.I.ItisdifferentfromthewayofA.I.Ithaveselflearnjustlikeabrianofhuman.Itiseasytohearspeakreadandwrite.Itdesignedinaeasyprocess.Itonlyhavedata.Anditwillaaccordingtothisinformationtoselflearn.Anditwaswhywechoosetheartificialneuralnetwork.KeywordartificialneuralnetworkA.Iselflearn.233234235∑∑∑+−=ijjJjjIJiXXWXEθ)2/1(njiiniijnJXXWfX=−=∑+)(1θ∑=−iiniIJiXW0θ∑−iiniIJiXW0θ∑−iiniIJiXW0θ23601001000000100104321DCBA≠xjxijixXX⋅∑∑∑≠1≠yixyxiixXX⋅∑∑∑≠2−∑∑NXixix237≠≠()1,1,−+≠+⋅⋅∑∑∑IYiyxixyxyixXXXdist()∑∑∑∑∑∑∑∑∑∑∑−+≠≠≠+••+−+•+•=xiyiyixixyxyxixiyixixjxixjxixyxiXXXdistXXXXXDNCBAE1,1,22222()()()()()()()()()NCijijdistDCyxjiBjiyxAxixyxiyjW⋅−=−++⋅⋅−−−⋅⋅−−⋅⋅−=θδδδδδδ1,1,,1,,1,()deltaba=,δ()ijxyCAWxiyi≠=−−=,()ijxyCBWxiyi=≠−−=,()1=−−⋅−=ijCdistDxyxiyiW()CWxiyi−=⋅−−−=00000000000000000022ddddddddWxzxzxyxyxxxxxwxwDCCCCCCCCCCCCCCCCBBAAAB238=()01111=−⋅=+++∑∑netXXWnetnxinxixinyjyjxiyjnxiθ()001=+netnxi00000000000000004321\zyxw∑∑−•=yjixjyXjyixWixnet]][[]][[]][][][[]][[θ239⋅−−−=00000000000000000022ddddddddWxzxzxyxyxxxxxwxwDCCCCCCCCCCCCCCCCBBAAAB==4444444444444444θ22≤220121101221011210WXyzwxyz01001000000100102242640100100000010010112122−−=+−−=+⋅++−=21221222212222net240VisualBasic6.0