%權值更新函數
%輸入參數：net，權值待更新的卷積神經網絡；opts，神經網絡訓練的相關參數
%算法流程：先更新卷積層的參數，再更新全連接層參數
function?net?=?cnnapplygrads（net?opts）%使用梯度
????%特征抽取層（卷機降采樣）的權重更新
????for?l?=?2?:?numel（net.layers）%從第二層開始
????????if?strcmp（net.layers{l}.type?‘c‘）%對于每個卷積層
????????????for?j?=?1?:?numel（net.layers{l}.a）%枚舉該層的每個輸出
????????????????%枚舉所有卷積核net.layers{l}.k{ii}{j}
????????????????for?ii?=?1?:?numel（net.layers{l?-?1}.a）%枚舉上層的每個輸出，普通的權值更新的公式：W_new?=?W_old?-?alpha?*?de/dW（誤差對權值導數）??
????????????????????net.layers{l}.k{ii}{j}?=?net.layers{l}.k{ii}{j}?-?opts.alpha?*?net.layers{l}.dk{ii}{j};
????????????????end
????????????end
????????????%修正
????????????net.layers{l}.b{j}?=?net.layers{l}.b{j}?-?opts.alpha?*?net.layers{l}.db{j};
????????end
????end
????%單層感知機的權重更新
????net.ffW?=?net.ffW?-?opts.alpha?*?net.dffW;
????net.ffb?=

?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----
?????文件?????????238??2017-12-28?16:33??CNN1維代碼\all_y.mat
?????文件?????????991??2016-11-27?21:15??CNN1維代碼\cnnapplygrads.m
?????文件????????4169??2016-11-27?21:29??CNN1維代碼\cnnbp.m
?????文件????????3847??2016-11-27?21:22??CNN1維代碼\cnnff.m
?????文件????????3590??2013-01-05?05:45??CNN1維代碼\cnnnumgradcheck.m
?????文件????????6196??2016-11-27?21:09??CNN1維代碼\cnnsetup.m
?????文件????????1038??2017-12-28?16:31??CNN1維代碼\cnntest.m
?????文件????????3442??2018-06-30?21:54??CNN1維代碼\cnntrain.m
?????文件?????????816??2018-04-08?15:39??CNN1維代碼\DATAtq.m
?????文件????????1958??2013-01-05?05:45??CNN1維代碼\expand.m
?????文件??????????80??2013-01-05?05:45??CNN1維代碼\flipall.m
?????文件????????4869??2018-06-27?21:21??CNN1維代碼\main.m
?????文件??????????48??2013-01-05?05:45??CNN1維代碼\sigm.m
?????文件?????6889380??2017-12-28?16:33??CNN1維代碼\yall_x.mat
?????文件???????15456??2018-01-19?16:19??CNN1維代碼\調用函數流程.png
?????文件???????????0??2017-09-21?21:27??CNN1維代碼\輸出結果中的right：第一列分錯的樣本序號；第二列正確的分類；第三列錯分到的類。.txt
?????文件????????2618??2013-01-05?05:45??CNN1維代碼\util\allcomb.m
?????文件????????1958??2013-01-05?05:45??CNN1維代碼\util\expand.m
?????文件?????????208??2013-01-05?05:45??CNN1維代碼\util\flicker.m
?????文件??????????80??2013-01-05?05:45??CNN1維代碼\util\flipall.m
?????文件?????????543??2013-01-05?05:45??CNN1維代碼\util\fliplrf.m
?????文件?????????576??2013-01-05?05:45??CNN1維代碼\util\flipudf.m
?????文件?????????289??2013-01-05?05:45??CNN1維代碼\util\im2patches.m
?????文件????????1895??2013-01-05?05:45??CNN1維代碼\util\makeLMfilters.m
?????文件?????????242??2013-01-05?05:45??CNN1維代碼\util\patches2im.m
?????文件?????????283??2013-01-05?05:45??CNN1維代碼\util\randcorr.m
?????文件????????2083??2013-01-05?05:45??CNN1維代碼\util\randp.m
?????文件??????????49??2013-01-05?05:45??CNN1維代碼\util\rnd.m
?????文件??????????48??2013-01-05?05:45??CNN1維代碼\util\sigm.m
?????文件?????????126??2013-01-05?05:45??CNN1維代碼\util\sigmrnd.m
?????文件?????????256??2013-01-05?05:45??CNN1維代碼\util\softmax.m
............此處省略41個文件信息