官术网_书友最值得收藏!

Getting ready

To understand the impact of varying the optimizer on network accuracy, let's contrast the scenario laid out in previous sections (which was the Adam optimizer) with using a stochastic gradient descent optimizer in this section, while reusing the same MNIST training and test datasets that were scaled (the same data-preprocessing steps as those of step 1 and step 2 in the Scaling the dataset recipe):

model = Sequential()
model.add(Dense(1000, input_dim=784, activation='relu'))
model.add(Dense(10, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])
history = model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=100, batch_size=32, verbose=1)

Note that when we used the stochastic gradient descent optimizer in the preceding code, the final accuracy after 100 epochs is ~98% (the code to generate the plots in the following diagram remains the same as the code we used in step 8 of the Training a vanilla neural network recipe):

However, we should also note that the model achieved the high accuracy levels much more slowly when compared to the model that used Adam optimization.

主站蜘蛛池模板: 香港| 调兵山市| 乌兰县| 遂宁市| 石嘴山市| 唐山市| 瓦房店市| 赤壁市| 万安县| 抚顺市| 南宫市| 昂仁县| 龙门县| 安溪县| 西青区| 晋宁县| 清丰县| 平顺县| 个旧市| 淮安市| 翼城县| 武冈市| 沾益县| 辽宁省| 怀柔区| 广元市| 建阳市| 盐边县| 云和县| 雷山县| 鹤岗市| 陆川县| 安宁市| 华坪县| 健康| 蛟河市| 赤峰市| 壶关县| 车致| 安康市| 西贡区|