4.keras-交叉熵的介紹和應用

keras-交叉熵的介紹和應用

1.載入數據以及預處理

import numpy as np
from keras.datasets import mnist from keras.utils import np_utils from keras.models import Sequential from keras.layers import * from keras.optimizers import SGD import os import tensorflow as tf # 載入數據 (x_train,y_train),(x_test,y_test) = mnist.load_data() # 預處理 # 將（60000，28，28）轉化爲（600000，784），好輸入展開層 x_train = x_train.reshape(x_train.shape[0],-1)/255.0 x_test= x_test.reshape(x_test.shape[0],-1)/255.0 # 將輸出轉化爲one_hot編碼 y_train = np_utils.to_categorical(y_train,num_classes=10) y_test = np_utils.to_categorical(y_test,num_classes=10)

2.建立網絡打印訓練結果

# 建立網絡
model = Sequential([
    # 輸入784輸出10個
    Dense(units=10,input_dim=784,bias_initializer='one',activation='softmax') ]) # 編譯 # 自定義優化器 sgd = SGD(lr=0.1) model.compile(optimizer=sgd,
　　　　　　　　# 運用交叉熵 loss='categorical_crossentropy', metrics=['accuracy']) model.fit(x_train,y_train,batch_size=32,epochs=10,validation_split=0.2) # 評估模型 loss,acc = model.evaluate(x_test,y_test,) print('\ntest loss',loss) print('test acc',acc)

out:

Epoch 1/10

32/48000 [..............................] - ETA: 2:43 - loss: 2.2593 - acc: 0.1562
1792/48000 [>.............................] - ETA: 4s - loss: 1.2642 - acc: 0.6579

......

......

Epoch 10/10

47456/48000 [============================>.] - ETA: 0s - loss: 0.2712 - acc: 0.9241
48000/48000 [==============================] - 2s 41us/step - loss: 0.2716 - acc: 0.9240 - val_loss: 0.2748 - val_acc: 0.9240

 

32/10000 [..............................] - ETA: 0s
2976/10000 [=======>......................] - ETA: 0s
6656/10000 [==================>...........] - ETA: 0s
10000/10000 [==============================] - 0s 17us/step

test loss 0.2802182431191206test acc 0.9205