Keras\u662f\u7531\u7eafpython\u7f16\u5199\u7684\u57fa\u4e8etheano\/tensorflow\u7684\u6df1\u5ea6\u5b66\u4e60\u6846\u67b6\u3002<\/p>\n

Keras\u662f\u4e00\u4e2a\u9ad8\u5c42\u795e\u7ecf\u7f51\u7edcAPI\uff0c\u652f\u6301\u5feb\u901f\u5b9e\u9a8c\uff0c\u80fd\u591f\u628a\u4f60\u7684idea\u8fc5\u901f\u8f6c\u6362\u4e3a\u7ed3\u679c\uff0c\u5982\u679c\u6709\u5982\u4e0b\u9700\u6c42\uff0c\u53ef\u4ee5\u4f18\u5148\u9009\u62e9Keras\uff1a<\/p>\n

a\uff09\u7b80\u6613\u548c\u5feb\u901f\u7684\u539f\u578b\u8bbe\u8ba1\uff08keras\u5177\u6709\u9ad8\u5ea6\u6a21\u5757\u5316\uff0c\u6781\u7b80\uff0c\u548c\u53ef\u6269\u5145\u7279\u6027\uff09<\/p>\n

b\uff09\u652f\u6301CNN\u548cRNN\uff0c\u6216\u4e8c\u8005\u7684\u7ed3\u5408<\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0\u00a0<\/span>c\uff09\u65e0\u7f1dCPU\u548cGPU\u5207\u6362<\/p>\n

a\uff09\u7528\u6237\u53cb\u597d\uff1aKeras\u662f\u4e3a\u4eba\u7c7b\u800c\u4e0d\u662f\u5929\u9876\u661f\u4eba\u8bbe\u8ba1\u7684API\u3002\u7528\u6237\u7684\u4f7f\u7528\u4f53\u9a8c\u59cb\u7ec8\u662f\u6211\u4eec\u8003\u8651\u7684\u9996\u8981\u548c\u4e2d\u5fc3\u5185\u5bb9\u3002Keras\u9075\u5faa\u51cf\u5c11\u8ba4\u77e5\u56f0\u96be\u7684\u6700\u4f73\u5b9e\u8df5\uff1aKeras\u63d0\u4f9b\u4e00\u81f4\u800c\u7b80\u6d01\u7684API\uff0c \u80fd\u591f\u6781\u5927\u51cf\u5c11\u4e00\u822c\u5e94\u7528\u4e0b\u7528\u6237\u7684\u5de5\u4f5c\u91cf\uff0c\u540c\u65f6\uff0cKeras\u63d0\u4f9b\u6e05\u6670\u548c\u5177\u6709\u5b9e\u8df5\u610f\u4e49\u7684bug\u53cd\u9988\u3002<\/p>\n

b\uff09\u6a21\u5757\u6027\uff1a\u6a21\u578b\u53ef\u7406\u89e3\u4e3a\u4e00\u4e2a\u5c42\u7684\u5e8f\u5217\u6216\u6570\u636e\u7684\u8fd0\u7b97\u56fe\uff0c\u5b8c\u5168\u53ef\u914d\u7f6e\u7684\u6a21\u5757\u53ef\u4ee5\u7528\u6700\u5c11\u7684\u4ee3\u4ef7\u81ea\u7531\u7ec4\u5408\u5728\u4e00\u8d77\u3002\u5177\u4f53\u800c\u8a00\uff0c\u7f51\u7edc\u5c42\u3001\u635f\u5931\u51fd\u6570\u3001\u4f18\u5316\u5668\u3001\u521d\u59cb\u5316\u7b56\u7565\u3001\u6fc0\u6d3b\u51fd\u6570\u3001\u6b63\u5219\u5316\u65b9\u6cd5\u90fd\u662f\u72ec\u7acb\u7684\u6a21\u5757\uff0c\u4f60\u53ef\u4ee5\u4f7f\u7528\u5b83\u4eec\u6765\u6784\u5efa\u81ea\u5df1\u7684\u6a21\u578b\u3002<\/p>\n

c\uff09\u6613\u6269\u5c55\u6027\uff1a\u6dfb\u52a0\u65b0\u6a21\u5757\u8d85\u7ea7\u5bb9\u6613\uff0c\u53ea\u9700\u8981\u4eff\u7167\u73b0\u6709\u7684\u6a21\u5757\u7f16\u5199\u65b0\u7684\u7c7b\u6216\u51fd\u6570\u5373\u53ef\u3002\u521b\u5efa\u65b0\u6a21\u5757\u7684\u4fbf\u5229\u6027\u4f7f\u5f97Keras\u66f4\u9002\u5408\u4e8e\u5148\u8fdb\u7684\u7814\u7a76\u5de5\u4f5c\u3002<\/p>\n

d\uff09\u4e0ePython\u534f\u4f5c\uff1aKeras\u6ca1\u6709\u5355\u72ec\u7684\u6a21\u578b\u914d\u7f6e\u6587\u4ef6\u7c7b\u578b\uff08\u4f5c\u4e3a\u5bf9\u6bd4\uff0ccaffe\u6709\uff09\uff0c\u6a21\u578b\u7531python\u4ee3\u7801\u63cf\u8ff0\uff0c\u4f7f\u5176\u66f4\u7d27\u51d1\u548c\u66f4\u6613debug\uff0c\u5e76\u63d0\u4f9b\u4e86\u6269\u5c55\u7684\u4fbf\u5229\u6027\u3002<\/p>\n

3.\u4f7f\u7528Keras\u642d\u5efa\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc<\/p>\n $\"\"$ \n

Keras\u7684\u5e95\u5c42\u5e93\u4f7f\u7528Theano\u6216TensorFlow\uff0c\u8fd9\u4e24\u4e2a\u5e93\u4e5f\u79f0\u4e3aKeras\u7684\u540e\u7aef\u3002\u65e0\u8bba\u662fTheano\u8fd8\u662fTensorFlow\uff0c\u90fd\u662f\u4e00\u4e2a\u201c\u7b26\u53f7\u5f0f\u201d\u7684\u5e93\u3002\u7b26\u53f7\u8ba1\u7b97\u9996\u5148\u5b9a\u4e49\u5404\u79cd\u53d8\u91cf\uff0c\u7136\u540e\u5efa\u7acb\u4e00\u4e2a\u201c\u8ba1\u7b97\u56fe\u201d,\u8ba1\u7b97\u56fe\u89c4\u5b9a\u4e86\u5404\u4e2a\u53d8\u91cf\u4e4b\u95f4\u7684\u8ba1\u7b97\u5173\u7cfb\u3002<\/p>\n

\u7b26\u53f7\u8ba1\u7b97\u4e5f\u53eb\u6570\u636e\u6d41\u56fe\uff0c\u5176\u8fc7\u7a0b\u5982\u4e0b(gif\u56fe\u4e0d\u597d\u6253\u5f00\uff0c\u6240\u4ee5\u7528\u4e86\u9759\u6001\u56fe\uff0c\u6570\u636e\u662f\u6309\u56fe\u4e2d\u9ed1\u8272\u5e26\u7bad\u5934\u7684\u7ebf\u6d41\u52a8\u7684)\uff1a<\/p>\n $\"\"$ \n

\u5f20\u91cf(tensor)\uff0c\u53ef\u4ee5\u770b\u4f5c\u662f\u5411\u91cf\u3001\u77e9\u9635\u7684\u81ea\u7136\u63a8\u5e7f\uff0c\u7528\u6765\u8868\u793a\u5e7f\u6cdb\u7684\u6570\u636e\u7c7b\u578b\u3002\u5f20\u91cf\u7684\u9636\u6570\u4e5f\u53eb\u7ef4\u5ea6\u3002<\/p>\n

1\u9636\u5f20\u91cf,\u5373\u5411\u91cf,\u4e00\u7ec4\u6709\u5e8f\u6392\u5217\u7684\u6570<\/p>\n

2\u9636\u5f20\u91cf,\u5373\u77e9\u9635,\u4e00\u7ec4\u5411\u91cf\u6709\u5e8f\u7684\u6392\u5217\u8d77\u6765<\/p>\n

3\u9636\u5f20\u91cf\uff0c\u5373\u7acb\u65b9\u4f53\uff0c\u4e00\u7ec4\u77e9\u9635\u4e0a\u4e0b\u6392\u5217\u8d77\u6765<\/p>\n

\u5047\u5982\u6709\u4e00\u4e2a10\u957f\u5ea6\u7684\u5217\u8868\uff0c\u90a3\u4e48\u6211\u4eec\u6a2a\u5411\u770b\u670910\u4e2a\u6570\u5b57\uff0c\u4e5f\u53ef\u4ee5\u53eb\u505a10\u7ef4\u5ea6\uff0c\u7eb5\u5411\u770b\u53ea\u80fd\u770b\u52301\u4e2a\u6570\u5b57\uff0c\u90a3\u4e48\u5c31\u53eb1\u7ef4\u5ea6\u3002\u6ce8\u610f\u8fd9\u4e2a\u533a\u522b\u6709\u52a9\u4e8e\u7406\u89e3Keras\u6216\u8005\u795e\u7ecf\u7f51\u7edc\u4e2d\u8ba1\u7b97\u65f6\u51fa\u73b0\u7684\u7ef4\u5ea6\u95ee\u9898\u3002<\/p>\n

\u76ee\u524d\u4e3b\u8981\u6709\u4e24\u79cd\u65b9\u5f0f\u6765\u8868\u793a\u5f20\u91cf\uff1a\na) th\u6a21\u5f0f\u6216channels_first\u6a21\u5f0f\uff0cTheano\u548ccaffe\u4f7f\u7528\u6b64\u6a21\u5f0f\u3002\nb\uff09tf\u6a21\u5f0f\u6216channels_last\u6a21\u5f0f\uff0cTensorFlow\u4f7f\u7528\u6b64\u6a21\u5f0f\u3002\n\u4e0b\u9762\u4e3e\u4f8b\u8bf4\u660e\u4e24\u79cd\u6a21\u5f0f\u7684\u533a\u522b\uff1a\n\u5bf9\u4e8e100\u5f20RGB3\u901a\u9053\u768416\u00d732\uff08\u9ad8\u4e3a16\u5bbd\u4e3a32\uff09\u5f69\u8272\u56fe\uff0c\nth\u8868\u793a\u65b9\u5f0f\uff1a\uff08100,3,16,32\uff09\ntf\u8868\u793a\u65b9\u5f0f\uff1a\uff08100,16,32,3\uff09\n\u552f\u4e00\u7684\u533a\u522b\u5c31\u662f\u8868\u793a\u901a\u9053\u4e2a\u65703\u7684\u4f4d\u7f6e\u4e0d\u4e00\u6837\u3002<\/p>\n

Keras\u6709\u4e24\u79cd\u7c7b\u578b\u7684\u6a21\u578b\uff0c\u5e8f\u8d2f\u6a21\u578b\uff08Sequential\uff09\u548c\u51fd\u6570\u5f0f\u6a21\u578b\uff08Model\uff09\uff0c\u51fd\u6570\u5f0f\u6a21\u578b\u5e94\u7528\u66f4\u4e3a\u5e7f\u6cdb\uff0c\u5e8f\u8d2f\u6a21\u578b\u662f\u51fd\u6570\u5f0f\u6a21\u578b\u7684\u4e00\u79cd\u7279\u6b8a\u60c5\u51b5\u3002\na\uff09\u5e8f\u8d2f\u6a21\u578b\uff08Sequential):\u5355\u8f93\u5165\u5355\u8f93\u51fa\uff0c\u4e00\u6761\u8def\u901a\u5230\u5e95\uff0c\u5c42\u4e0e\u5c42\u4e4b\u95f4\u53ea\u6709\u76f8\u90bb\u5173\u7cfb\uff0c\u6ca1\u6709\u8de8\u5c42\u8fde\u63a5\u3002\u8fd9\u79cd\u6a21\u578b\u7f16\u8bd1\u901f\u5ea6\u5feb\uff0c\u64cd\u4f5c\u4e5f\u6bd4\u8f83\u7b80\u5355\nb\uff09\u51fd\u6570\u5f0f\u6a21\u578b\uff08Model\uff09\uff1a\u591a\u8f93\u5165\u591a\u8f93\u51fa\uff0c\u5c42\u4e0e\u5c42\u4e4b\u95f4\u4efb\u610f\u8fde\u63a5\u3002\u8fd9\u79cd\u6a21\u578b\u7f16\u8bd1\u901f\u5ea6\u6162\u3002<\/p>\n

\u8fd9\u91cc\u4e5f\u91c7\u7528\u4ecb\u7ecd\u795e\u7ecf\u7f51\u7edc\u65f6\u5e38\u7528\u7684\u4e00\u4e2a\u4f8b\u5b50\uff1a\u624b\u5199\u6570\u5b57\u7684\u8bc6\u522b\u3002<\/p>\n

\u5728\u5199\u4ee3\u7801\u4e4b\u524d\uff0c\u57fa\u4e8e\u8fd9\u4e2a\u4f8b\u5b50\u4ecb\u7ecd\u4e00\u4e9b\u6982\u5ff5\uff0c\u65b9\u4fbf\u5927\u5bb6\u7406\u89e3\u3002<\/p>\n

PS\uff1a\u53ef\u80fd\u662f\u7248\u672c\u5dee\u5f02\u7684\u95ee\u9898\uff0c\u5b98\u7f51\u4e2d\u7684\u53c2\u6570\u548c\u793a\u4f8b\u4e2d\u7684\u53c2\u6570\u662f\u4e0d\u4e00\u6837\u7684\uff0c\u5b98\u7f51\u4e2d\u7ed9\u51fa\u7684\u53c2\u6570\u5c11\uff0c\u5e76\u4e14\u6709\u4e9b\u53c2\u6570\u652f\u6301\uff0c\u6709\u4e9b\u4e0d\u652f\u6301\u3002\u6240\u4ee5\u6b64\u4f8b\u5b50\u53bb\u6389\u4e86\u4e0d\u652f\u6301\u7684\u53c2\u6570\uff0c\u5e76\u4e14\u53ea\u4ecb\u7ecd\u672c\u4f8b\u4e2d\u7528\u5230\u7684\u53c2\u6570\u3002<\/p>\n

a\uff09Dense\u5c42\u5c5e\u4e8e\u7f51\u7edc\u5c42–>\u5e38\u7528\u5c42\u4e2d\u7684\u4e00\u4e2a\u5c42<\/p>\n

b\uff09 500\u8868\u793a\u8f93\u51fa\u7684\u7ef4\u5ea6\uff0c\u5b8c\u6574\u7684\u8f93\u51fa\u8868\u793a\uff1a(*,500)\uff1a\u5373\u8f93\u51fa\u4efb\u610f\u4e2a500\u7ef4\u7684\u6570\u636e\u6d41\u3002\u4f46\u662f\u5728\u53c2\u6570\u4e2d\u53ea\u5199\u7ef4\u5ea6\u5c31\u53ef\u4ee5\u4e86\uff0c\u6bd4\u8f83\u5177\u4f53\u8f93\u51fa\u591a\u5c11\u4e2a\u662f\u6709\u8f93\u5165\u786e\u5b9a\u7684\u3002\u6362\u4e2a\u8bf4\u6cd5\uff0cDense\u7684\u8f93\u51fa\u5176\u5b9e\u662f\u4e2aN\u00d7500\u7684\u77e9\u9635\u3002<\/p>\n

c\uff09input_shape(784,) \u8868\u793a\u8f93\u5165\u7ef4\u5ea6\u662f784(28\u00d728\uff0c\u540e\u9762\u5177\u4f53\u4ecb\u7ecd\u4e3a\u4ec0\u4e48)\uff0c\u5b8c\u6574\u7684\u8f93\u5165\u8868\u793a\uff1a(*,784)\uff1a\u5373\u8f93\u5165N\u4e2a784\u7ef4\u5ea6\u7684\u6570\u636e<\/p>\n

\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u6bcf\u6b21\u66f4\u65b0\u53c2\u6570\u65f6\u968f\u673a\u65ad\u5f00\u4e00\u5b9a\u767e\u5206\u6bd4\uff08rate\uff09\u7684\u8f93\u5165\u795e\u7ecf\u5143\uff0c\u9632\u6b62\u8fc7\u62df\u5408\u3002<\/p>\n

\u6570\u636e\u96c6\u5305\u62ec60000\u5f2028\u00d728\u7684\u8bad\u7ec3\u96c6\u548c10000\u5f2028\u00d728\u7684\u6d4b\u8bd5\u96c6\u53ca\u5176\u5bf9\u5e94\u7684\u76ee\u6807\u6570\u5b57\u3002\u5982\u679c\u5b8c\u5168\u6309\u7167\u4e0a\u8ff0\u6570\u636e\u683c\u5f0f\u8868\u8ff0\uff0c\u4ee5tensorflow\u4f5c\u4e3a\u540e\u7aef\u5e94\u8be5\u662f\uff0860000,28,28,3\uff09\uff0c\u56e0\u4e3a\u793a\u4f8b\u4e2d\u91c7\u7528\u4e86mnist.load_data()\u83b7\u53d6\u6570\u636e\u96c6\uff0c\u6240\u4ee5\u5df2\u7ecf\u5224\u65ad\u4f7f\u7528\u4e86tensorflow\u4f5c\u4e3a\u540e\u7aef\uff0c\u56e0\u6b64\u6570\u636e\u96c6\u5c31\u53d8\u6210\u4e86(60000,28,28),\u90a3\u4e48input_shape(784,)\u5e94\u8be5\u662finput_shape(28,28\uff0c)\u624d\u5bf9\uff0c\u4f46\u662f\u5728\u8fd9\u4e2a\u793a\u4f8b\u4e2d\u8fd9\u4e48\u5199\u662f\u4e0d\u5bf9\u7684\uff0c\u9700\u8981\u8f6c\u6362\u6210(60000,784),\u624d\u53ef\u4ee5\u3002\u4e3a\u4ec0\u4e48\u9700\u8981\u8f6c\u6362\u5462\uff1f<\/p>\n $\"\"$ \n

\u5982\u4e0a\u56fe\uff0c\u8bad\u7ec3\u96c6(60000,28,28)\u4f5c\u4e3a\u8f93\u5165\uff0c\u5c31\u76f8\u5f53\u4e8e\u4e00\u4e2a\u7acb\u65b9\u4f53\uff0c\u800c\u8f93\u5165\u5c42\u4ece\u5f53\u524d\u89d2\u5ea6\u770b\u5c31\u662f\u4e00\u4e2a\u5e73\u9762\uff0c\u7acb\u65b9\u4f53\u7684\u6570\u636e\u6d41\u600e\u4e48\u8fdb\u5165\u5e73\u9762\u7684\u8f93\u5165\u5c42\u8fdb\u884c\u8ba1\u7b97\u5462\uff1f\u6240\u4ee5\u9700\u8981\u8fdb\u884c\u9ec4\u8272\u7bad\u5934\u6240\u793a\u7684\u53d8\u6362\uff0c\u7136\u540e\u624d\u8fdb\u5165\u8f93\u5165\u5c42\u8fdb\u884c\u540e\u7eed\u8ba1\u7b97\u3002\u81f3\u4e8e\u4ece28*28\u53d8\u6362\u6210784\u4e4b\u540e\u8f93\u5165\u5c42\u5982\u4f55\u5904\u7406\uff0c\u5c31\u4e0d\u9700\u8981\u6211\u4eec\u5173\u5fc3\u4e86\u3002(\u559c\u6b22\u94bb\u7814\u7684\u540c\u5b66\u53ef\u4ee5\u53bb\u7814\u7a76\u4e0b\u6e90\u4ee3\u7801)\u3002<\/p>\n

\u5e76\u4e14\uff0cKeras\u4e2d\u8f93\u5165\u591a\u4e3a(nb_samples, input_dim)\u7684\u5f62\u5f0f\uff1a\u5373(\u6837\u672c\u6570\u91cf\uff0c\u8f93\u5165\u7ef4\u5ea6)\u3002<\/p>\n

from keras.models import Sequential  \nfrom keras.layers.core import Dense, Dropout, Activation  \nfrom keras.optimizers import SGD  \nfrom keras.datasets import mnist  \nimport numpy \n'''\n    \u7b2c\u4e00\u6b65\uff1a\u9009\u62e9\u6a21\u578b\n'''\nmodel = Sequential()\n'''\n   \u7b2c\u4e8c\u6b65\uff1a\u6784\u5efa\u7f51\u7edc\u5c42\n'''\nmodel.add(Dense(500,input_shape=(784,))) # \u8f93\u5165\u5c42\uff0c28*28=784  \nmodel.add(Activation('tanh')) # \u6fc0\u6d3b\u51fd\u6570\u662ftanh  \nmodel.add(Dropout(0.5)) # \u91c7\u752850%\u7684dropout\n\nmodel.add(Dense(500)) # \u9690\u85cf\u5c42\u8282\u70b9500\u4e2a  \nmodel.add(Activation('tanh'))  \nmodel.add(Dropout(0.5))\n\nmodel.add(Dense(10)) # \u8f93\u51fa\u7ed3\u679c\u662f10\u4e2a\u7c7b\u522b\uff0c\u6240\u4ee5\u7ef4\u5ea6\u662f10  \nmodel.add(Activation('softmax')) # \u6700\u540e\u4e00\u5c42\u7528softmax\u4f5c\u4e3a\u6fc0\u6d3b\u51fd\u6570\n\n'''\n   \u7b2c\u4e09\u6b65\uff1a\u7f16\u8bd1\n'''\nsgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) # \u4f18\u5316\u51fd\u6570\uff0c\u8bbe\u5b9a\u5b66\u4e60\u7387\uff08lr\uff09\u7b49\u53c2\u6570  \nmodel.compile(loss='categorical_crossentropy', optimizer=sgd, class_mode='categorical') # \u4f7f\u7528\u4ea4\u53c9\u71b5\u4f5c\u4e3aloss\u51fd\u6570\n\n'''\n   \u7b2c\u56db\u6b65\uff1a\u8bad\u7ec3\n   .fit\u7684\u4e00\u4e9b\u53c2\u6570\n   batch_size\uff1a\u5bf9\u603b\u7684\u6837\u672c\u6570\u8fdb\u884c\u5206\u7ec4\uff0c\u6bcf\u7ec4\u5305\u542b\u7684\u6837\u672c\u6570\u91cf\n   epochs \uff1a\u8bad\u7ec3\u6b21\u6570\n   shuffle\uff1a\u662f\u5426\u628a\u6570\u636e\u968f\u673a\u6253\u4e71\u4e4b\u540e\u518d\u8fdb\u884c\u8bad\u7ec3\n   validation_split\uff1a\u62ff\u51fa\u767e\u5206\u4e4b\u591a\u5c11\u7528\u6765\u505a\u4ea4\u53c9\u9a8c\u8bc1\n   verbose\uff1a\u5c4f\u663e\u6a21\u5f0f 0\uff1a\u4e0d\u8f93\u51fa  1\uff1a\u8f93\u51fa\u8fdb\u5ea6  2\uff1a\u8f93\u51fa\u6bcf\u6b21\u7684\u8bad\u7ec3\u7ed3\u679c\n'''\n(X_train, y_train), (X_test, y_test) = mnist.load_data() # \u4f7f\u7528Keras\u81ea\u5e26\u7684mnist\u5de5\u5177\u8bfb\u53d6\u6570\u636e\uff08\u7b2c\u4e00\u6b21\u9700\u8981\u8054\u7f51\uff09\n# \u7531\u4e8emist\u7684\u8f93\u5165\u6570\u636e\u7ef4\u5ea6\u662f(num, 28, 28)\uff0c\u8fd9\u91cc\u9700\u8981\u628a\u540e\u9762\u7684\u7ef4\u5ea6\u76f4\u63a5\u62fc\u8d77\u6765\u53d8\u6210784\u7ef4  \nX_train = X_train.reshape(X_train.shape[0], X_train.shape[1] * X_train.shape[2]) \nX_test = X_test.reshape(X_test.shape[0], X_test.shape[1] * X_test.shape[2])  \nY_train = (numpy.arange(10) == y_train[:, None]).astype(int) \nY_test = (numpy.arange(10) == y_test[:, None]).astype(int)\n\nmodel.fit(X_train,Y_train,batch_size=200,epochs=50,shuffle=True,verbose=0,validation_split=0.3)\nmodel.evaluate(X_test, Y_test, batch_size=200, verbose=0)\n\n'''\n    \u7b2c\u4e94\u6b65\uff1a\u8f93\u51fa\n'''\nprint(\"test set\")\nscores = model.evaluate(X_test,Y_test,batch_size=200,verbose=0)\nprint(\"\")\nprint(\"The test loss is %f\" % scores)\nresult = model.predict(X_test,batch_size=200,verbose=0)\n\nresult_max = numpy.argmax(result, axis = 1)\ntest_max = numpy.argmax(Y_test, axis = 1)\n\nresult_bool = numpy.equal(result_max, test_max)\ntrue_num = numpy.sum(result_bool)\nprint(\"\")\nprint(\"The accuracy of the model is %f\" % (true_num\/len(result_bool)))<\/pre>\n<\/div>\n <\/p>","protected":false},"excerpt":{"rendered":"
1.\u5173\u4e8eKeras 1\uff09\u7b80\u4ecb Keras\u662f\u7531\u7eafpython\u7f16\u5199\u7684\u57fa\u4e8etheano\/tensorflow\u7684\u6df1\u5ea6\u5b66\u4e60\u6846\u67b6\u3002 Keras\u662f\u4e00\u4e2a\u9ad8\u5c42\u795e\u7ecf\u7f51\u7edcAPI\uff0c\u652f\u6301\u5feb\u901f\u5b9e\u9a8c\uff0c\u80fd\u591f\u628a\u4f60\u7684idea\u8fc5\u901f\u8f6c\u6362\u4e3a\u7ed3...<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[274],"tags":[500],"class_list":["post-1942","post","type-post","status-publish","format-standard","hentry","category-python","tag-keras"],"_links":{"self":[{"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/posts\/1942","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/comments?post=1942"}],"version-history":[{"count":3,"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/posts\/1942\/revisions"}],"predecessor-version":[{"id":1954,"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/posts\/1942\/revisions\/1954"}],"wp:attachment":[{"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/media?parent=1942"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/categories?post=1942"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/nick.txtcc.com\/index.php\/wp-json\/wp\/v2\/tags?post=1942"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}