{"id":433,"date":"2014-09-14T03:25:39","date_gmt":"2014-09-13T18:25:39","guid":{"rendered":"https:\/\/www.dogrow.net\/nnet\/?p=433"},"modified":"2024-01-20T02:08:42","modified_gmt":"2024-01-19T17:08:42","slug":"blog26","status":"publish","type":"post","link":"https:\/\/www.dogrow.net\/nnet\/blog26\/","title":{"rendered":"(26) cuda-convnet2\u3067\u30ed\u30b8\u30b9\u30c6\u30a3\u30c3\u30af\u56de\u5e30\u306a\u4e8c\u5024\u5206\u985e\u5668\u3092\u4f5c\u3063\u3066\u307f\u308b"},"content":{"rendered":"

cuda-convnet2<\/span> \u3067\u306f\u3001\u65b0\u305f\u306b cross-entropy cost layer<\/span> \u304c\u8ffd\u52a0\u3055\u308c\u305f\u3002<\/p>\n

1. \u3053\u308c\u3092\u4f7f\u3046\u3068\u4f55\u304c\u3046\u308c\u3057\u3044\uff1f<\/h1>\n

softmax \u3092\u7528\u3044\u305f\u591a\u5024\u5206\u985e\u5668\u3067\u306f\u3001\u51fa\u529b\u30e6\u30cb\u30c3\u30c8\u9593\u306e\u76f8\u5bfe\u7684\u306a\u5927\u5c0f\u95a2\u4fc2\u3092\u5b66\u7fd2\u3055\u305b\u3066
\n\u300c\u6700\u5927\u51fa\u529b\u30e6\u30cb\u30c3\u30c8\u756a\u53f7\uff1d\u4ed8\u4e0e\u3059\u308b\u30e9\u30d9\u30eb<\/span>\u300d \u3068\u3057\u3066\u4f7f\u7528\u3057\u305f\u3002<\/p>\n

\u3053\u308c\u306b\u5bfe\u3057\u30662\u5024\u5206\u985e\u5668\u3067\u306f\u3001\u4e00\u3064\u306e\u51fa\u529b\u30e6\u30cb\u30c3\u30c8\u304c \u300c\u3042\u308b\u6761\u4ef6\u306b\u5bfe\u3057\u3066 Yes or No<\/span>\u300d \u3060\u3051\u3092\u5b66\u7fd2\u3059\u308b\u3002
\n\u3053\u308c\u3092\u5fdc\u7528\u3057\u3001\u8907\u6570\u500b\u306e\u51fa\u529b\u30e6\u30cb\u30c3\u30c8\u306b\u72ec\u7acb\u3057\u305f\u6761\u4ef6\u5224\u5b9a\u30ed\u30b8\u30c3\u30af\u3092\u5b66\u7fd2\u3055\u305b\u308c\u3070\u3001\u4e00\u3064\u306e\u5165\u529b\u30c7\u30fc\u30bf\u306b\u5bfe\u3057\u3066\u8907\u6570\u7a2e\u985e\u306e\u5224\u5b9a\u7d50\u679c\u304c\u51fa\u529b\u53ef\u80fd\u306a\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u3092\u4f5c\u308c\u308b\u3002<\/p>\n

2. \u5b9f\u9a13\u306e\u6e96\u5099<\/h1>\n

(1) \u5b9f\u9a13\u6761\u4ef6<\/h3>\n

\u30d6\u30ed\u30b0\u306e\u30c6\u30fc\u30de\u304cMNIST\u306a\u306e\u3067\u3001\u3053\u3053\u3067\u3082MNIST\u3092\u4f7f\u3063\u3066\u5b9f\u9a13\u3059\u308b\u3002
\n1) \u5b66\u7fd2\u30fb\u30c6\u30b9\u30c8\u753b\u50cf\u306fMNIST\u6570\u5b57\u753b\u50cf\u30c7\u30fc\u30bf\u3092\u4f7f\u7528\u3059\u308b\u3002
\n2) \u30d9\u30fc\u30b9\u3068\u306a\u308b\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u69cb\u6210\u306f (25) cuda-convnet2\u3067MNIST\u81ea\u52d5\u8a8d\u8b58(\u305d\u306e1)<\/a> \u3067\u4f7f\u7528\u3057\u305f\u3082\u306e\u3002
\n3) \u51fa\u529b\u5c64\u3092 detection cross-entropy cost layer<\/span> \u306b\u5909\u66f4\u3059\u308b\u3002
\n4) \u51fa\u529b\u30e6\u30cb\u30c3\u30c8\u65703\u500b\u3092\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u5b9a\u7fa9\u3059\u308b\u3002
\n\n
\u51fa\u529b\u30e6\u30cb\u30c3\u30c8No.1<\/td>\u5947\u6570\u30d5\u30e9\u30b0<\/td><\/tr>
\u51fa\u529b\u30e6\u30cb\u30c3\u30c8No.2<\/td>\u5076\u6570\u30d5\u30e9\u30b0<\/td><\/tr>
\u51fa\u529b\u30e6\u30cb\u30c3\u30c8No.3<\/td>3\u306e\u500d\u6570\u30d5\u30e9\u30b0<\/td><\/tr>\n<\/table><\/p>\n

\u3053\u306e\u30eb\u30fc\u30eb\u306b\u5f93\u3046\u3068\u3001\u5165\u529b\u3057\u305f MNIST\u6570\u5b57\u753b\u50cf 0\uff5e9 \u306b\u5bfe\u3059\u308b\u51fa\u529b\u306f\u6b21\u306e\u3088\u3046\u306b\u306a\u308b\u3002
\n\n
\u6570\u5b57<\/td>0<\/td>1<\/td>2<\/td>3<\/td>4<\/td>5<\/td>6<\/td>7<\/td>8<\/td>9<\/td><\/tr>
\u51fa\u529b\u5024(bin)<\/td>010<\/td>100<\/td>010<\/td>101<\/td>010<\/td>100<\/td>011<\/td>100<\/td>010<\/td>101<\/td><\/tr>\n<\/table><\/p>\n

(2) \u5b66\u7fd2\u30fb\u30c6\u30b9\u30c8\u30c7\u30fc\u30bf\u4f5c\u308a<\/h3>\n

cuda-convnet\u7528MNIST\u30c7\u30fc\u30bf\u3092\u4f5c\u308b(\u305d\u306e2)<\/a>, (\u305d\u306e3)<\/a> \u3067\u4f5c\u6210\u3057\u305f cuda-convnet\u5165\u529b\u30c7\u30fc\u30bf\u4f5c\u6210\u30c4\u30fc\u30eb\u306b\u5c11\u3005\u624b\u3092\u5165\u308c\u308b\u5fc5\u8981\u3042\u308a\u3002\u6539\u9020\u7b87\u6240\u306f1\u70b9\u3060\u3051\u3002
\n1) \u6559\u5e2b\u30c7\u30fc\u30bf\u306f\u30e9\u30d9\u30eb\u756a\u53f7\u3067\u306f\u306a\u304f\u3001\u4e0a\u8a18\u306e\u51fa\u529b\u5c64\u306e\u51fa\u529b\u5024(3\u6b21\u5143\u60c5\u5831)\u3068\u3059\u308b\u3002<\/p>\n

\u4f8b) \u4eca\u307e\u3067\u306e\u591a\u5024\u5206\u985e\u3067\u306f data_batch_N <\/span>\u30d5\u30a1\u30a4\u30eb\u4e2d\u306e\u6559\u5e2b\u30c7\u30fc\u30bf\u3092\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u8a2d\u5b9a\u3057\u3066\u3044\u305f\u3002 (\u5024\u306f\u9069\u5f53)<\/p>\n

labels = [0,3,2,6,1,3]<\/pre>\n
\u4eca\u56de\u306e2\u5024\u5206\u985e\u5668\u3067\u306f\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u8a2d\u5b9a\u3059\u308b\u3002(\u5024\u306f\u9069\u5f53)<\/p>\n
labels = [[0,1,0],[1,0,0],[1,1,0],[1,0,1]]\r\n<\/pre>\n
(3) NN\u5b9a\u7fa9\u30d5\u30a1\u30a4\u30eb\u4f5c\u308a<\/h3>\n
layers-MNIST-tri.cfg<\/span>
\n\u6700\u5f8c\u306e\u30d5\u30eb\u63a5\u7d9a\u5c64\u306e neuron \u3092 logistic \u306b\u3057\u306a\u3044\u3068\u5b9f\u884c\u6642\u306b\u30a8\u30e9\u30fc\u304c\u51fa\u308b\u3002<\/p>\n
[data]\r\ntype=data\r\ndataIdx=0\r\n\r\n[labels]\r\ntype=data\r\ndataIdx=1\r\n\r\n[conv1]\r\ntype=conv\r\ninputs=data\r\nchannels=1\r\nfilters=32\r\npadding=0\r\nstride=1\r\nfilterSize=5\r\nneuron=tanh[1,1]\r\ninitW=0.0001\r\nsumWidth=4\r\nsharedBiases=1\r\ngpu=0\r\n\r\n[pool1]\r\ntype=pool\r\npool=max\r\ninputs=conv1\r\nstart=0\r\nsizeX=2\r\nstride=2\r\noutputsX=0\r\nchannels=32\r\n\r\n[conv2]\r\ntype=conv\r\ninputs=pool1\r\nfilters=32\r\npadding=0\r\nstride=1\r\nfilterSize=5\r\nchannels=32\r\nneuron=tanh[1,1]\r\ninitW=0.01\r\nsumWidth=2\r\nsharedBiases=1\r\n\r\n[pool2]\r\ntype=pool\r\npool=avg\r\ninputs=conv2\r\nstart=0\r\nsizeX=2\r\nstride=2\r\noutputsX=0\r\nchannels=32\r\n\r\n[fcOut]\r\ntype=fc\r\noutputs=3\r\ninputs=pool2\r\ninitW=0.01\r\ninitB=0.1\r\nneuron=logistic<\/span>\r\n\r\n[dce]\r\ntype=cost.dce<\/span>\r\ninputs=labels,fcOut\r\ngpu=0\r\n<\/pre>\n
layer-params-MNIST-tri.cfg<\/span><\/p>\n
[conv1]\r\nepsW=0.01\r\nepsB=0.01\r\nmomW=0.9\r\nmomB=0.9\r\nwc=0.0001\r\n\r\n[conv2]\r\nepsW=0.01\r\nepsB=0.01\r\nmomW=0.9\r\nmomB=0.9\r\nwc=0.0001\r\n\r\n[fcOut]\r\nepsW=0.01\r\nepsB=0.01\r\nmomW=0.9\r\nmomB=0.9\r\nwc=0.0001\r\n\r\n[dce]\r\ncoeff=1\r\ntopk=1\r\n<\/pre>\n
3. \u5b9f\u884c\u7d50\u679c<\/h1>\n
10epochs \u6d41\u3057\u3066\u307f\u305f\u3002
\n\u4eca\u307e\u3067\u3088\u308a\u3082\u30ed\u30b0\u51fa\u529b\u3055\u308c\u308b\u60c5\u5831\u91cf\u304c\u591a\u3044\u3088\u3046\u3060\u3002<\/p>\n
2014\u5e74  9\u6708 14\u65e5 \u65e5\u66dc\u65e5 02:21:24 JST\r\nInitialized data layer 'data', producing 784 outputs\r\nInitialized data layer 'labels', producing 3 outputs\r\nInitialized convolutional layer 'conv1' on GPUs 0, producing 24x24 32-channel output\r\nInitialized max-pooling layer 'pool1' on GPUs 0, producing 12x12 32-channel output\r\nInitialized convolutional layer 'conv2' on GPUs 0, producing 8x8 32-channel output\r\nInitialized avg-pooling layer 'pool2' on GPUs 0, producing 4x4 32-channel output\r\nInitialized fully-connected layer 'fcOut' on GPUs 0, producing 3 outputs\r\nInitialized detection cross-entropy cost 'dce' on GPUs 0\r\nInitialized neuron layer 'fcOut_neuron' on GPUs 0, producing 3 outputs\r\nInitialized neuron layer 'conv2_neuron' on GPUs 0, producing 2048 outputs\r\nInitialized neuron layer 'conv1_neuron' on GPUs 0, producing 18432 outputs\r\nLayer conv2_neuron using acts from layer conv2\r\nLayer fcOut_neuron using acts from layer fcOut\r\nLayer conv1_neuron using acts from layer conv1\r\n=========================\r\nImporting cudaconvnet._ConvNet C++ module\r\nFwd terminal: dce\r\nfound bwd terminal conv1[0] in passIdx=0\r\n=========================\r\nTraining ConvNet\r\nAdd PCA noise to color channels with given scale                        : 0   [DEFAULT]\r\nCheck gradients and quit?                                               : 0   [DEFAULT]\r\nConserve GPU memory (slower)?                                           : 0   [DEFAULT]\r\nConvert given conv layers to unshared local                             :\r\nCropped DP: crop size (0 = don't crop)                                  : 28\r\nCropped DP: test on multiple patches?                                   : 0   [DEFAULT]\r\nData batch range: testing                                               : 7-7\r\nData batch range: training                                              : 1-6\r\nData path                                                               : \/home\/user\/cuda\/cuda-convnet2\/data\/MNIST-tri\/\r\nData provider                                                           : DCE\r\nForce save before quitting                                              : 0   [DEFAULT]\r\nGPU override                                                            : 0\r\nLayer definition file                                                   : \/home\/user\/cuda\/cuda-convnet2\/config\/MNIST\/layers-MNIST-tri.cfg\r\nLayer file path prefix                                                  :     [DEFAULT]\r\nLayer parameter file                                                    : \/home\/user\/cuda\/cuda-convnet2\/config\/MNIST\/layer-params-MNIST-tri.cfg\r\nLoad file                                                               :     [DEFAULT]\r\nLogreg cost layer name (for --test-out)                                 :     [DEFAULT]\r\nMinibatch size                                                          : 128 [DEFAULT]\r\nNumber of epochs                                                        : 10\r\nOutput test case predictions to given path                              :     [DEFAULT]\r\nSave file override                                                      :\r\nSave path                                                               : \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/\r\nSubtract this scalar from image (-1 = don't)                            : -1  [DEFAULT]\r\nTest and quit?                                                          : 0   [DEFAULT]\r\nTest on one batch at a time?                                            : 1   [DEFAULT]\r\nTesting frequency                                                       : 6\r\nUnshare weight matrices in given layers                                 :\r\nWrite test data features from given layer                               :     [DEFAULT]\r\nWrite test data features to this path (to be used with --write-features):     [DEFAULT]\r\n=========================\r\nRunning on CUDA device(s) 0\r\nCurrent time: Sun Sep 14 02:21:26 2014\r\nSaving checkpoints to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n=========================\r\n1.1 (0.00%)... dce:  (crossent) 1.199113, (err) 0.183633, (Godd) 0.813920, 0.890335, (Geven) 0.848251, 0.845842, (Gtri) 0.706835, 0.683478 (0.563 sec)\r\n1.2 (1.67%)... dce:  (crossent) 0.766782, (err) 0.103000, (Godd) 0.919650, 0.914490, (Geven) 0.908736, 0.915131, (Gtri) 0.832205, 0.823325 (0.253 sec)\r\n1.3 (3.33%)... dce:  (crossent) 0.567953, (err) 0.068433, (Godd) 0.950119, 0.945626, (Geven) 0.944995, 0.949025, (Gtri) 0.883268, 0.860500 (0.256 sec)\r\n1.4 (5.00%)... dce:  (crossent) 0.443588, (err) 0.052900, (Godd) 0.963820, 0.960780, (Geven) 0.958771, 0.963053, (Gtri) 0.897314, 0.897088 (0.251 sec)\r\n1.5 (6.67%)... dce:  (crossent) 0.410039, (err) 0.049000, (Godd) 0.964023, 0.958687, (Geven) 0.958937, 0.964177, (Gtri) 0.914705, 0.908954 (0.256 sec)\r\n1.6 (8.33%)... dce:  (crossent) 0.349523, (err) 0.040433, (Godd) 0.967851, 0.963834, (Geven) 0.963680, 0.966802, (Gtri) 0.938540, 0.928083\r\n======================Test output======================\r\ndce:  (crossent) 0.317917, (err) 0.036367, (Godd) 0.973128, 0.970635, (Geven) 0.969439, 0.972391, (Gtri) 0.944646, 0.922921\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.317917, (err) 0.036367, (Godd) 0.973128, 0.970635, (Geven) 0.969439, 0.972391, (Gtri) 0.944646, 0.922921\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 8.155207e-02 [7.599560e-04] [9.318661e-03]\r\nLayer 'conv1' biases: 2.710742e-03 [2.686247e-05]\r\nLayer 'conv2' weights[0]: 2.011198e-02 [1.423615e-04] [7.078444e-03]\r\nLayer 'conv2' biases: 2.185383e-02 [2.591967e-04]\r\nLayer 'fcOut' weights[0]: 8.837998e-02 [4.182930e-04] [4.732893e-03]\r\nLayer 'fcOut' biases: 1.263399e-01 [7.117551e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.380 sec)\r\n2.1 (10.00%)... dce:  (crossent) 0.309526, (err) 0.035000, (Godd) 0.973669, 0.970020, (Geven) 0.969685, 0.973225, (Gtri) 0.941206, 0.938634 (0.262 sec)\r\n2.2 (11.67%)... dce:  (crossent) 0.305727, (err) 0.035700, (Godd) 0.973188, 0.969046, (Geven) 0.968035, 0.971641, (Gtri) 0.942188, 0.938213 (0.257 sec)\r\n2.3 (13.33%)... dce:  (crossent) 0.301048, (err) 0.033967, (Godd) 0.976615, 0.970843, (Geven) 0.971296, 0.975833, (Gtri) 0.943435, 0.931514 (0.256 sec)\r\n2.4 (15.00%)... dce:  (crossent) 0.269323, (err) 0.030567, (Godd) 0.978870, 0.976941, (Geven) 0.976287, 0.977873, (Gtri) 0.942836, 0.939759 (0.255 sec)\r\n2.5 (16.67%)... dce:  (crossent) 0.262750, (err) 0.030367, (Godd) 0.980865, 0.972722, (Geven) 0.972105, 0.980368, (Gtri) 0.945326, 0.943662 (0.258 sec)\r\n2.6 (18.33%)... dce:  (crossent) 0.232449, (err) 0.027100, (Godd) 0.979010, 0.977075, (Geven) 0.975994, 0.979352, (Gtri) 0.957322, 0.948595\r\n======================Test output======================\r\ndce:  (crossent) 0.245786, (err) 0.028500, (Godd) 0.987382, 0.971620, (Geven) 0.971611, 0.986602, (Gtri) 0.976171, 0.911044\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.245786, (err) 0.028500, (Godd) 0.987382, 0.971620, (Geven) 0.971611, 0.986602, (Gtri) 0.976171, 0.911044\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.103154e-01 [6.913404e-04] [6.266943e-03]\r\nLayer 'conv1' biases: 3.618342e-03 [1.293126e-05]\r\nLayer 'conv2' weights[0]: 2.478795e-02 [1.239718e-04] [5.001291e-03]\r\nLayer 'conv2' biases: 2.868914e-02 [1.545805e-04]\r\nLayer 'fcOut' weights[0]: 1.136545e-01 [2.959493e-04] [2.603940e-03]\r\nLayer 'fcOut' biases: 1.387306e-01 [3.311247e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.391 sec)\r\n3.1 (20.00%)... dce:  (crossent) 0.224097, (err) 0.025033, (Godd) 0.981969, 0.977515, (Geven) 0.976974, 0.981136, (Gtri) 0.958230, 0.957516 (0.274 sec)\r\n3.2 (21.67%)... dce:  (crossent) 0.230526, (err) 0.026400, (Godd) 0.979814, 0.976591, (Geven) 0.974845, 0.978679, (Gtri) 0.957795, 0.957320 (0.274 sec)\r\n3.3 (23.33%)... dce:  (crossent) 0.228185, (err) 0.025833, (Godd) 0.981050, 0.979117, (Geven) 0.977935, 0.981113, (Gtri) 0.956699, 0.949204 (0.268 sec)\r\n3.4 (25.00%)... dce:  (crossent) 0.213223, (err) 0.023333, (Godd) 0.984202, 0.982263, (Geven) 0.981969, 0.983963, (Gtri) 0.954568, 0.954568 (0.266 sec)\r\n3.5 (26.67%)... dce:  (crossent) 0.208459, (err) 0.023700, (Godd) 0.983148, 0.980233, (Geven) 0.979625, 0.982797, (Gtri) 0.960953, 0.953219 (0.266 sec)\r\n3.6 (28.33%)... dce:  (crossent) 0.185518, (err) 0.021233, (Godd) 0.983759, 0.981621, (Geven) 0.981018, 0.983401, (Gtri) 0.965438, 0.962016\r\n======================Test output======================\r\ndce:  (crossent) 0.204320, (err) 0.023067, (Godd) 0.990608, 0.976941, (Geven) 0.976972, 0.990459, (Gtri) 0.983051, 0.923427\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.204320, (err) 0.023067, (Godd) 0.990608, 0.976941, (Geven) 0.976972, 0.990459, (Gtri) 0.983051, 0.923427\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.302679e-01 [6.364945e-04] [4.886041e-03]\r\nLayer 'conv1' biases: 3.856979e-03 [8.572064e-06]\r\nLayer 'conv2' weights[0]: 2.778544e-02 [1.004645e-04] [3.615725e-03]\r\nLayer 'conv2' biases: 3.361420e-02 [9.831827e-05]\r\nLayer 'fcOut' weights[0]: 1.289217e-01 [2.037354e-04] [1.580304e-03]\r\nLayer 'fcOut' biases: 1.446893e-01 [1.287783e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.397 sec)\r\n4.1 (30.00%)... dce:  (crossent) 0.188140, (err) 0.021433, (Godd) 0.983564, 0.979684, (Geven) 0.978998, 0.983367, (Gtri) 0.967380, 0.965217 (0.270 sec)\r\n4.2 (31.67%)... dce:  (crossent) 0.199008, (err) 0.022367, (Godd) 0.981395, 0.979687, (Geven) 0.978917, 0.980335, (Gtri) 0.968337, 0.963772 (0.274 sec)\r\n4.3 (33.33%)... dce:  (crossent) 0.194720, (err) 0.021700, (Godd) 0.984011, 0.982072, (Geven) 0.981557, 0.983550, (Gtri) 0.962494, 0.959818 (0.273 sec)\r\n4.4 (35.00%)... dce:  (crossent) 0.178280, (err) 0.019867, (Godd) 0.987352, 0.984628, (Geven) 0.984015, 0.987211, (Gtri) 0.962003, 0.959588 (0.274 sec)\r\n4.5 (36.67%)... dce:  (crossent) 0.181196, (err) 0.020500, (Godd) 0.986097, 0.981419, (Geven) 0.981273, 0.986238, (Gtri) 0.967529, 0.959256 (0.266 sec)\r\n4.6 (38.33%)... dce:  (crossent) 0.163559, (err) 0.018567, (Godd) 0.985550, 0.983992, (Geven) 0.983633, 0.985425, (Gtri) 0.967636, 0.969106\r\n======================Test output======================\r\ndce:  (crossent) 0.190747, (err) 0.021300, (Godd) 0.989081, 0.981868, (Geven) 0.981452, 0.988226, (Gtri) 0.986279, 0.926459\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.190747, (err) 0.021300, (Godd) 0.989081, 0.981868, (Geven) 0.981452, 0.988226, (Gtri) 0.986279, 0.926459\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.452356e-01 [5.611114e-04] [3.863456e-03]\r\nLayer 'conv1' biases: 4.199057e-03 [9.355912e-06]\r\nLayer 'conv2' weights[0]: 3.022003e-02 [9.920518e-05] [3.282762e-03]\r\nLayer 'conv2' biases: 3.737093e-02 [1.362906e-04]\r\nLayer 'fcOut' weights[0]: 1.403167e-01 [1.990263e-04] [1.418408e-03]\r\nLayer 'fcOut' biases: 1.483610e-01 [2.851311e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.400 sec)\r\n5.1 (40.00%)... dce:  (crossent) 0.172653, (err) 0.019367, (Godd) 0.984570, 0.981657, (Geven) 0.981201, 0.984584, (Gtri) 0.974661, 0.965217 (0.269 sec)\r\n5.2 (41.67%)... dce:  (crossent) 0.177714, (err) 0.019533, (Godd) 0.986410, 0.982975, (Geven) 0.982663, 0.985510, (Gtri) 0.967629, 0.964268 (0.269 sec)\r\n5.3 (43.33%)... dce:  (crossent) 0.170255, (err) 0.018900, (Godd) 0.986751, 0.983058, (Geven) 0.982598, 0.986190, (Gtri) 0.968568, 0.965631 (0.277 sec)\r\n5.4 (45.00%)... dce:  (crossent) 0.163894, (err) 0.018067, (Godd) 0.986985, 0.986401, (Geven) 0.985804, 0.986805, (Gtri) 0.966801, 0.964859 (0.286 sec)\r\n5.5 (46.67%)... dce:  (crossent) 0.162201, (err) 0.018133, (Godd) 0.987108, 0.983791, (Geven) 0.983458, 0.986642, (Gtri) 0.971176, 0.966046 (0.271 sec)\r\n5.6 (48.33%)... dce:  (crossent) 0.148354, (err) 0.016267, (Godd) 0.987532, 0.986166, (Geven) 0.985651, 0.987247, (Gtri) 0.971660, 0.972398\r\n======================Test output======================\r\ndce:  (crossent) 0.172621, (err) 0.019200, (Godd) 0.989861, 0.981277, (Geven) 0.981672, 0.989444, (Gtri) 0.985964, 0.940864\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.172621, (err) 0.019200, (Godd) 0.989861, 0.981277, (Geven) 0.981672, 0.989444, (Gtri) 0.985964, 0.940864\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.587927e-01 [5.413409e-04] [3.409105e-03]\r\nLayer 'conv1' biases: 4.576801e-03 [1.049395e-05]\r\nLayer 'conv2' weights[0]: 3.219037e-02 [9.470111e-05] [2.941908e-03]\r\nLayer 'conv2' biases: 4.058194e-02 [1.346725e-04]\r\nLayer 'fcOut' weights[0]: 1.493520e-01 [1.707393e-04] [1.143201e-03]\r\nLayer 'fcOut' biases: 1.511034e-01 [2.599937e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.400 sec)\r\n6.1 (50.00%)... dce:  (crossent) 0.157575, (err) 0.017633, (Godd) 0.985004, 0.984615, (Geven) 0.983783, 0.984381, (Gtri) 0.974807, 0.970932 (0.268 sec)\r\n6.2 (51.67%)... dce:  (crossent) 0.163027, (err) 0.017433, (Godd) 0.988528, 0.983556, (Geven) 0.982485, 0.986959, (Gtri) 0.972865, 0.969727 (0.268 sec)\r\n6.3 (53.33%)... dce:  (crossent) 0.151389, (err) 0.015667, (Godd) 0.988946, 0.986998, (Geven) 0.986426, 0.988830, (Gtri) 0.973838, 0.968916 (0.269 sec)\r\n6.4 (55.00%)... dce:  (crossent) 0.150452, (err) 0.015900, (Godd) 0.988952, 0.987978, (Geven) 0.987830, 0.988632, (Gtri) 0.971270, 0.967369 (0.277 sec)\r\n6.5 (56.67%)... dce:  (crossent) 0.149220, (err) 0.017100, (Godd) 0.986931, 0.985175, (Geven) 0.984848, 0.986642, (Gtri) 0.974411, 0.967304 (0.268 sec)\r\n6.6 (58.33%)... dce:  (crossent) 0.141361, (err) 0.016100, (Godd) 0.986169, 0.986364, (Geven) 0.986829, 0.985830, (Gtri) 0.973178, 0.973917\r\n======================Test output======================\r\ndce:  (crossent) 0.166090, (err) 0.019333, (Godd) 0.984634, 0.985022, (Geven) 0.984169, 0.984369, (Gtri) 0.985000, 0.945919\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.166090, (err) 0.019333, (Godd) 0.984634, 0.985022, (Geven) 0.984169, 0.984369, (Gtri) 0.985000, 0.945919\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.691544e-01 [5.411348e-04] [3.199059e-03]\r\nLayer 'conv1' biases: 4.764370e-03 [1.533124e-05]\r\nLayer 'conv2' weights[0]: 3.393728e-02 [1.179916e-04] [3.476754e-03]\r\nLayer 'conv2' biases: 4.355372e-02 [1.882927e-04]\r\nLayer 'fcOut' weights[0]: 1.571101e-01 [2.250832e-04] [1.432646e-03]\r\nLayer 'fcOut' biases: 1.538871e-01 [4.014182e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.395 sec)\r\n7.1 (60.00%)... dce:  (crossent) 0.146865, (err) 0.016633, (Godd) 0.986553, 0.984024, (Geven) 0.983607, 0.985801, (Gtri) 0.977295, 0.973168 (0.270 sec)\r\n7.2 (61.67%)... dce:  (crossent) 0.148073, (err) 0.015967, (Godd) 0.988176, 0.986264, (Geven) 0.985133, 0.987580, (Gtri) 0.975330, 0.971216 (0.269 sec)\r\n7.3 (63.33%)... dce:  (crossent) 0.139719, (err) 0.014200, (Godd) 0.989154, 0.988180, (Geven) 0.987827, 0.988830, (Gtri) 0.976679, 0.973717 (0.275 sec)\r\n7.4 (65.00%)... dce:  (crossent) 0.137491, (err) 0.014633, (Godd) 0.989927, 0.987781, (Geven) 0.987047, 0.990053, (Gtri) 0.974817, 0.971637 (0.280 sec)\r\n7.5 (66.67%)... dce:  (crossent) 0.138026, (err) 0.015767, (Godd) 0.989094, 0.985966, (Geven) 0.985870, 0.988464, (Gtri) 0.974482, 0.970070 (0.283 sec)\r\n7.6 (68.33%)... dce:  (crossent) 0.131545, (err) 0.014533, (Godd) 0.988142, 0.988142, (Geven) 0.987647, 0.987247, (Gtri) 0.974249, 0.977209\r\n======================Test output======================\r\ndce:  (crossent) 0.147773, (err) 0.017033, (Godd) 0.988528, 0.985022, (Geven) 0.984827, 0.988226, (Gtri) 0.981832, 0.956027\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.147773, (err) 0.017033, (Godd) 0.988528, 0.985022, (Geven) 0.984827, 0.988226, (Gtri) 0.981832, 0.956027\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.788537e-01 [6.211860e-04] [3.473151e-03]\r\nLayer 'conv1' biases: 5.401216e-03 [1.268923e-05]\r\nLayer 'conv2' weights[0]: 3.544452e-02 [1.119315e-04] [3.157935e-03]\r\nLayer 'conv2' biases: 4.565623e-02 [1.653222e-04]\r\nLayer 'fcOut' weights[0]: 1.637556e-01 [2.176586e-04] [1.329167e-03]\r\nLayer 'fcOut' biases: 1.556429e-01 [3.749391e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.415 sec)\r\n8.1 (70.00%)... dce:  (crossent) 0.135743, (err) 0.015233, (Godd) 0.987367, 0.986588, (Geven) 0.986226, 0.987627, (Gtri) 0.978505, 0.972671 (0.272 sec)\r\n8.2 (71.67%)... dce:  (crossent) 0.142047, (err) 0.015733, (Godd) 0.988936, 0.985684, (Geven) 0.985145, 0.988408, (Gtri) 0.975579, 0.971464 (0.279 sec)\r\n8.3 (73.33%)... dce:  (crossent) 0.132628, (err) 0.015767, (Godd) 0.988544, 0.986013, (Geven) 0.985616, 0.988018, (Gtri) 0.974398, 0.971443 (0.282 sec)\r\n8.4 (75.00%)... dce:  (crossent) 0.134742, (err) 0.014500, (Godd) 0.990713, 0.988175, (Geven) 0.987654, 0.990662, (Gtri) 0.975246, 0.969127 (0.273 sec)\r\n8.5 (76.67%)... dce:  (crossent) 0.128446, (err) 0.014467, (Godd) 0.990871, 0.986954, (Geven) 0.987094, 0.990690, (Gtri) 0.974773, 0.971831 (0.280 sec)\r\n8.6 (78.33%)... dce:  (crossent) 0.122649, (err) 0.014300, (Godd) 0.988338, 0.988142, (Geven) 0.987664, 0.988664, (Gtri) 0.976408, 0.974677\r\n======================Test output======================\r\ndce:  (crossent) 0.151681, (err) 0.018233, (Godd) 0.984077, 0.986598, (Geven) 0.986156, 0.983354, (Gtri) 0.979323, 0.957544\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.151681, (err) 0.018233, (Godd) 0.984077, 0.986598, (Geven) 0.986156, 0.983354, (Gtri) 0.979323, 0.957544\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.873097e-01 [5.005563e-04] [2.672345e-03]\r\nLayer 'conv1' biases: 5.523826e-03 [1.317529e-05]\r\nLayer 'conv2' weights[0]: 3.681655e-02 [9.820487e-05] [2.667411e-03]\r\nLayer 'conv2' biases: 4.807946e-02 [1.357576e-04]\r\nLayer 'fcOut' weights[0]: 1.698663e-01 [1.870220e-04] [1.100995e-03]\r\nLayer 'fcOut' biases: 1.567201e-01 [2.885009e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.418 sec)\r\n9.1 (80.00%)... dce:  (crossent) 0.129455, (err) 0.014500, (Godd) 0.988145, 0.986391, (Geven) 0.986232, 0.988032, (Gtri) 0.978120, 0.977391 (0.269 sec)\r\n9.2 (81.67%)... dce:  (crossent) 0.138982, (err) 0.015867, (Godd) 0.989122, 0.985104, (Geven) 0.984130, 0.988408, (Gtri) 0.975124, 0.972705 (0.278 sec)\r\n9.3 (83.33%)... dce:  (crossent) 0.130423, (err) 0.014233, (Godd) 0.988760, 0.987786, (Geven) 0.987419, 0.988221, (Gtri) 0.978664, 0.973717 (0.281 sec)\r\n9.4 (85.00%)... dce:  (crossent) 0.125232, (err) 0.013933, (Godd) 0.990521, 0.988569, (Geven) 0.988450, 0.990256, (Gtri) 0.973427, 0.974649 (0.278 sec)\r\n9.5 (86.67%)... dce:  (crossent) 0.126634, (err) 0.015000, (Godd) 0.989501, 0.987349, (Geven) 0.987278, 0.989476, (Gtri) 0.975696, 0.969316 (0.269 sec)\r\n9.6 (88.33%)... dce:  (crossent) 0.119767, (err) 0.013300, (Godd) 0.990493, 0.988340, (Geven) 0.988083, 0.990283, (Gtri) 0.977665, 0.975437\r\n======================Test output======================\r\ndce:  (crossent) 0.143420, (err) 0.015733, (Godd) 0.987016, 0.988766, (Geven) 0.988611, 0.986805, (Gtri) 0.982656, 0.959313\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.143420, (err) 0.015733, (Godd) 0.987016, 0.988766, (Geven) 0.988611, 0.986805, (Gtri) 0.982656, 0.959313\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 1.955977e-01 [5.039437e-04] [2.576430e-03]\r\nLayer 'conv1' biases: 5.744199e-03 [1.012264e-05]\r\nLayer 'conv2' weights[0]: 3.810624e-02 [1.096704e-04] [2.878017e-03]\r\nLayer 'conv2' biases: 5.125387e-02 [1.558181e-04]\r\nLayer 'fcOut' weights[0]: 1.754434e-01 [2.125424e-04] [1.211459e-03]\r\nLayer 'fcOut' biases: 1.569183e-01 [3.300385e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.410 sec)\r\n10.1 (90.00%)... dce:  (crossent) 0.124347, (err) 0.014267, (Godd) 0.987379, 0.987574, (Geven) 0.987024, 0.987424, (Gtri) 0.979811, 0.976646 (0.293 sec)\r\n10.2 (91.67%)... dce:  (crossent) 0.127718, (err) 0.014533, (Godd) 0.989333, 0.986845, (Geven) 0.985558, 0.988822, (Gtri) 0.978808, 0.974194 (0.268 sec)\r\n10.3 (93.33%)... dce:  (crossent) 0.121530, (err) 0.012833, (Godd) 0.991308, 0.988574, (Geven) 0.988252, 0.990861, (Gtri) 0.979436, 0.974981 (0.269 sec)\r\n10.4 (95.00%)... dce:  (crossent) 0.120518, (err) 0.012667, (Godd) 0.992289, 0.989160, (Geven) 0.988871, 0.992083, (Gtri) 0.977341, 0.974398 (0.270 sec)\r\n10.5 (96.67%)... dce:  (crossent) 0.121691, (err) 0.014867, (Godd) 0.988538, 0.988733, (Geven) 0.988266, 0.988666, (Gtri) 0.973784, 0.971579 (0.263 sec)\r\n10.6 (98.33%)... dce:  (crossent) 0.113142, (err) 0.013233, (Godd) 0.989715, 0.988933, (Geven) 0.988673, 0.989474, (Gtri) 0.977446, 0.976703\r\n======================Test output<\/span>======================\r\ndce:  (crossent) 0.136332, (err) 0.014800, (Godd) 0.985885, 0.991131, (Geven) 0.990814, 0.985384, (Gtri) 0.979770, 0.966894<\/span>\r\n----------------------Averages-------------------------\r\ndce:  (crossent) 0.136332, (err) 0.014800, (Godd) 0.985885, 0.991131, (Geven) 0.990814, 0.985384, (Gtri) 0.979770, 0.966894\r\n-------------------------------------------------------\r\nLayer 'conv1' weights[0]: 2.045221e-01 [5.905093e-04] [2.887264e-03]\r\nLayer 'conv1' biases: 6.253900e-03 [9.817431e-06]\r\nLayer 'conv2' weights[0]: 3.927734e-02 [9.667401e-05] [2.461318e-03]\r\nLayer 'conv2' biases: 5.389512e-02 [1.231133e-04]\r\nLayer 'fcOut' weights[0]: 1.806059e-01 [1.755370e-04] [9.719342e-04]\r\nLayer 'fcOut' biases: 1.589827e-01 [2.841916e-04]\r\n-------------------------------------------------------\r\nSaved checkpoint to \/home\/user\/cuda\/cuda-convnet2\/save\/MNIST\/ConvNet__2014-09-14_02.21.24\r\n======================================================= (0.395 sec)\r\n11.1 (100.00%)... dce:  (crossent) 0.118477, (err) 0.013900, (Godd) 0.988551, 0.987771, (Geven) 0.987231, 0.988032, (Gtri) 0.980289, 0.976149 (0.290 sec)\r\n2014\u5e74  9\u6708 14\u65e5 \u65e5\u66dc\u65e5 02:21:45 JST\r\n<\/pre>\n
Test output<\/span> \u306b\u8868\u793a\u3055\u308c\u3066\u3044\u308b\u306e\u306f\u4ee5\u4e0b\u306e\u3082\u306e\u3002
\n\n
(crossent) 0.136332<\/td>\u640d\u5931\u95a2\u6570\u51fa\u529b\u5024<\/td><\/tr>
(err) 0.014800<\/td>\u5168\u30c6\u30b9\u30c8\u30c7\u30fc\u30bf\u306b\u5bfe\u3059\u308b\u30a8\u30e9\u30fc\u7387 \u203b\u30a8\u30e9\u30fc\u51fa\u529b\u6570 \/ (\u30c6\u30b9\u30c8\u753b\u50cf\u679a\u6570 \u00d7 \u51fa\u529b\u30e6\u30cb\u30c3\u30c8\u6570)<\/td><\/tr>
(Godd) 0.985885, 0.991131<\/td>\u51fa\u529b\u30e6\u30cb\u30c3\u30c8#1\u306ePositive\u6b63\u89e3\u7387 \u30fb\u5de6\u5074\u306fPrecision<\/span>: TruePositive<\/span> \/ DeclaredPositive<\/span> \u30fb\u53f3\u5074\u306fRecall<\/span>: TruePositive<\/span> \/ Positive<\/span> \u203bbatches.meta\u3067Godd\u3068\u547d\u540d\u3057\u305f<\/td><\/tr>
(Geven) 0.990814, 0.985384<\/td>\u51fa\u529b\u30e6\u30cb\u30c3\u30c8No.2\u306e\u6b63\u89e3\u7387 \u203bbatches.meta\u3067Geven\u3068\u547d\u540d\u3057\u305f<\/td><\/tr>
(Gtri) 0.979770, 0.966894<\/td>\u51fa\u529b\u30e6\u30cb\u30c3\u30c8No.3\u306e\u6b63\u89e3\u7387 \u203bbatches.meta\u3067Gtri\u3068\u547d\u540d\u3057\u305f<\/td><\/tr>\n<\/table><\/p>\n\n
TruePositive<\/span><\/td>(\u6b63\u89e3\u5024, \u51fa\u529b\u5024) = (1, 1)<\/td><\/tr>
DeclaredPositive<\/span><\/td>\u51fa\u529b\u5024 = 1<\/td><\/tr>
Positive<\/span><\/td>\u6b63\u89e3\u5024 = 1<\/td><\/tr>\n<\/table>\n
======================Test output======================\r\ndce:  (crossent) 0.136332, (err) 0.014800, (Godd) 0.985885, 0.991131, (Geven) 0.990814, 0.985384, (Gtri) 0.979770, 0.966894\r\n<\/pre>\n
4. \u5ff5\u306e\u305f\u3081\u306b…<\/h1>\n
\u672c\u5f53\u306b\u5b66\u7fd2\u304c\u9032\u3093\u3067\u3044\u308b\u306e\u304b\u78ba\u8a8d\u3057\u3066\u307f\u305f\u304f\u306a\u3063\u305f\u306e\u3067\u3001\u4e0a\u8a18\u306e\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u5b9f\u884c\u4e2d\u306b
\nBinomialCrossEntropyCostLayer::fpropActs()<\/span> \u3067\u51fa\u529b\u5c64\u306e\u51fa\u529b\u5024\u3092\u30e2\u30cb\u30bf\u30ea\u30f3\u30b0\u3057\u3066\u307f\u305f\u3002(\u5404\u30d0\u30c3\u30c1\u30c7\u30fc\u30bf\u306e\u5148\u982d3\u500b\u3060\u3051)
\n\n
true label<\/td>\u6559\u5e2b\u30c7\u30fc\u30bf<\/td><\/tr>
prob.<\/td>\u51fa\u529b\u5c64\u306e\u751f\u51fa\u529b\u5024<\/td><\/tr>
prob.th.<\/td>\u95be\u50240.5\u3067\u4e8c\u5024\u5316\u6e08\u307f\u306e\u51fa\u529b\u5024<\/td><\/tr>\n<\/table><\/p>\n
1) epoch#1-batch#1<\/span><\/strong>
\n \u5f53\u7136\u3060\u3051\u308c\u3069\u3082\u307e\u3060\u307e\u3060\u5168\u7136\u30c0\u30e1\u30c0\u30e1…<\/span><\/span><\/p>\n
[src\/layer.cu][fpropActs][2117] true label(:,0) [1.000000<\/span>][0.000000][0.000000]\r\n[src\/layer.cu][fpropActs][2118] true label(:,1) [0.000000][1.000000<\/span>][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2119] true label(:,2) [0.000000][1.000000<\/span>][0.000000]\r\n\r\n[src\/layer.cu][fpropActs][2123] prob.     (:,0) [0.524935<\/span>][0.525479<\/span>][0.525543<\/span>]\r\n[src\/layer.cu][fpropActs][2124] prob.     (:,1) [0.525781<\/span>][0.525362<\/span>][0.524247<\/span>]\r\n[src\/layer.cu][fpropActs][2125] prob.     (:,2) [0.525054<\/span>][0.525753<\/span>][0.524198<\/span>]\r\n\u3000\u2193  threshold = 0.5<\/span>\r\n[src\/layer.cu][fpropActs][2141] prob.th.  (:,0) [1.000000<\/span>][1.000000<\/span>][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2142] prob.th.  (:,1) [1.000000<\/span>][1.000000<\/span>][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2143] prob.th.  (:,2) [1.000000<\/span>][1.000000<\/span>][1.000000<\/span>]\r\n<\/pre>\n
2) epoch#1-batch#2<\/span><\/strong>
\n\u307e\u30602\u30d0\u30c3\u30c1\u76ee(=\u5b66\u7fd2\u6e08\u307f\u30c7\u30fc\u30bf\u306f10,000\u500b) \u3060\u3051\u308c\u3069\u3082\u3053\u3053\u306f3\/3\u6b63\u89e3\uff01<\/p>\n
[src\/layer.cu][fpropActs][2117] true label(:,0) [1.000000<\/span>][0.000000][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2118] true label(:,1) [0.000000][1.000000<\/span>][0.000000]\r\n[src\/layer.cu][fpropActs][2119] true label(:,2) [1.000000<\/span>][0.000000][0.000000]\r\n\r\n[src\/layer.cu][fpropActs][2123] prob.     (:,0) [0.995084<\/span>][0.005174][0.962656<\/span>]\r\n[src\/layer.cu][fpropActs][2124] prob.     (:,1) [0.232831][0.785004<\/span>][0.025421]\r\n[src\/layer.cu][fpropActs][2125] prob.     (:,2) [0.992378<\/span>][0.008622][0.050015]\r\n\u3000\u2193  threshold = 0.5<\/span>\r\n[src\/layer.cu][fpropActs][2141] prob.th.  (:,0) [1.000000<\/span>][0.000000][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2142] prob.th.  (:,1) [0.000000][1.000000<\/span>][0.000000]\r\n[src\/layer.cu][fpropActs][2143] prob.th.  (:,2) [1.000000<\/span>][0.000000][0.000000]\r\n<\/pre>\n
3) epoch#1-batch#3<\/span><\/strong>
\n\u307e\u30603\u30d0\u30c3\u30c1\u76ee(=\u5b66\u7fd2\u6e08\u307f\u30c7\u30fc\u30bf\u306f20,000\u500b) \u60dc\u3057\u3044\u3051\u3069\u3061\u3087\u3063\u3068\u9593\u9055\u3048\u3066\u308b\u3002<\/p>\n
[src\/layer.cu][fpropActs][2117] true label(:,0) [1.000000<\/span>][0.000000][0.000000]\r\n[src\/layer.cu][fpropActs][2118] true label(:,1) [1.000000<\/span>][0.000000][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2119] true label(:,2) [1.000000<\/span>][0.000000][0.000000]\r\n\r\n[src\/layer.cu][fpropActs][2123] prob.     (:,0) [0.970718<\/span>][0.028501][0.343004]\r\n[src\/layer.cu][fpropActs][2124] prob.     (:,1) [0.854348<\/span>][0.155817][0.465876]\r\n[src\/layer.cu][fpropActs][2125] prob.     (:,2) [0.881519<\/span>][0.121382][0.047769]\r\n\u3000\u2193  threshold = 0.5<\/span>\r\n[src\/layer.cu][fpropActs][2141] prob.th.  (:,0) [1.000000<\/span>][0.000000][0.000000]\r\n[src\/layer.cu][fpropActs][2142] prob.th.  (:,1) [1.000000<\/span>][0.000000][0.000000]\r\n[src\/layer.cu][fpropActs][2143] prob.th.  (:,2) [1.000000<\/span>][0.000000][0.000000]\r\n<\/pre>\n
4) epoch#9-batch#2<\/span><\/strong>
\n\u3061\u3087\u3063\u3068\u9032\u3093\u30679epochs\u306e2\u30d0\u30c3\u30c1\u76ee(=\u5b66\u7fd2\u6e08\u307f\u30c7\u30fc\u30bf\u306f490,000\u500b) \u51fa\u529b\u5024\u306f\u307b\u307c\u307b\u307c\u6559\u5e2b\u30c7\u30fc\u30bf\u3068\u540c\u3058\u3060\uff01<\/p>\n
[src\/layer.cu][fpropActs][2117] true label(:,0) [1.000000<\/span>][0.000000][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2118] true label(:,1) [0.000000][1.000000<\/span>][0.000000]\r\n[src\/layer.cu][fpropActs][2119] true label(:,2) [1.000000<\/span>][0.000000][0.000000]\r\n\r\n[src\/layer.cu][fpropActs][2123] prob.     (:,0) [0.999963<\/span>][0.000038][0.999809<\/span>]\r\n[src\/layer.cu][fpropActs][2124] prob.     (:,1) [0.000016][0.999983<\/span>][0.000180]\r\n[src\/layer.cu][fpropActs][2125] prob.     (:,2) [0.999750<\/span>][0.000245][0.001855]\r\n\u3000\u2193  threshold = 0.5<\/span>\r\n[src\/layer.cu][fpropActs][2141] prob.th.  (:,0) [1.000000<\/span>][0.000000][1.000000<\/span>]\r\n[src\/layer.cu][fpropActs][2142] prob.th.  (:,1) [0.000000][1.000000<\/span>][0.000000]\r\n[src\/layer.cu][fpropActs][2143] prob.th.  (:,2) [1.000000<\/span>][0.000000][0.000000]\r\n<\/pre>\n
cuda-convnet2<\/span> \u3067\u3044\u308d\u3044\u308d\u3068\u904a\u3079\u308b\u5e45\u304c\u5e83\u304c\u3063\u305f(^o^)\/<\/p>\n
\n","protected":false},"excerpt":{"rendered":"
cuda-convnet2 \u3067\u306f\u3001\u65b0\u305f\u306b cross-entropy cost layer \u304c\u8ffd\u52a0\u3055\u308c\u305f\u3002 1. \u3053\u308c\u3092\u4f7f\u3046\u3068\u4f55\u304c\u3046\u308c\u3057\u3044\uff1f softmax \u3092\u7528\u3044\u305f\u591a\u5024\u5206\u985e\u5668\u3067\u306f\u3001\u51fa\u529b\u30e6\u30cb\u30c3\u30c8\u9593\u306e\u76f8\u5bfe\u7684\u306a\u5927\u5c0f\u95a2\u4fc2\u3092\u5b66\u7fd2\u2026 \u7d9a\u304d\u3092\u8aad\u3080 »<\/a><\/span><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8,10,11],"tags":[],"class_list":["post-433","post","type-post","status-publish","format-standard","hentry","category-cuda","category-cuda-convnet","category-cuda-convnet2"],"views":2823,"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/posts\/433","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/comments?post=433"}],"version-history":[{"count":86,"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/posts\/433\/revisions"}],"predecessor-version":[{"id":930,"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/posts\/433\/revisions\/930"}],"wp:attachment":[{"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/media?parent=433"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/categories?post=433"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.dogrow.net\/nnet\/wp-json\/wp\/v2\/tags?post=433"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}