tensorflow2.0 learn 2020-4-24

Author: budaLi

tenforlow_learn/demo-1.py

@@ -4,17 +4,32 @@
 # @Software: PyCharm
 # pip install --upgrade --ignore-installed tensorflow
 import matplotlib as mpl
+import os
 import matplotlib.pyplot as plt
 import numpy as np
 import tensorflow as tf
+import pandas as pd
 from tensorflow import keras
+from sklearn.preprocessing import StandardScaler
 
 fashion_mnist = keras.datasets.fashion_mnist
 # 共六万图片
 (x_train_all, y_train_all), (x_test, y_test) = fashion_mnist.load_data()
 x_valid, x_train = x_train_all[:5000], x_train_all[5000:]
 y_valid, y_train = y_train_all[:5000], y_train_all[5000:]
 
+print(np.max(x_train), np.min(x_train))
+
+# 做归一化
+scaler = StandardScaler()
+x_train_scaled = scaler.fit_transform(x_train.astype(np.float32).reshape(-1, 1)).reshape(-1, 28, 28)
+
+x_valid_scaled = scaler.transform(x_valid.astype(np.float32).reshape(-1, 1)).reshape(-1, 28, 28)
+
+x_test_scaled = scaler.transform(x_test.astype(np.float32).reshape(-1, 1)).reshape(-1, 28, 28)
+
+print(np.max(x_train_scaled), np.min(x_train_scaled))
+
 print(x_valid.shape, y_valid.shape)
 print(x_train.shape, y_train.shape)
 print(x_test.shape, y_test.shape)
@@ -35,15 +50,68 @@ def show_images(n_rows, n_cols, x_data, y_data, class_names):
     plt.figure(figsize=(n_cols * 1.4, n_rows * 1.6))
     for row in range(n_rows):
         for col in range(n_cols):
-            index = n_cols * n_rows + col
-            plt.subplot(n_rows, n_cols, index + 1)
+            # index 表示在n_rows行 n_cols列中的 哪个地方画图 取值范围为1，n_rows*n_cols
+            index = n_rows * n_cols - col - row * n_cols
+            plt.subplot(n_rows, n_cols, index)
             plt.imshow(x_data[index], cmap="binary", interpolation="nearest")
             plt.axis("off")
             plt.title(class_names[y_data[index]])
 
     plt.show()
 
 
+def train():
+    models = keras.models.Sequential()
+    # 输入28*28的图像数据 28*28的矩阵变为 1*726的一维向量
+    models.add(keras.layers.Flatten(input_shape=[28, 28]))
+    # 全连接层 relu激活函数
+    models.add(keras.layers.Dense(300, activation="relu"))
+
+    models.add(keras.layers.Dense(100, activation="relu"))
+
+    # softmax将向量变成概率分布
+    # x= [x1,x2,x3]
+    # y = [e^x1/sum,e^x2/sum,e^x3/sum],sum = e^x1+e^x2+e^x3
+    models.add(keras.layers.Dense(10, activation="softmax"))
+
+    models.compile(loss="sparse_categorical_crossentropy",
+                   optimizer="sgd",
+                   metrics=["accuracy"])
+
+    # 模型的层
+    print(models.layers)
+
+    # 可以看到模型的参数
+    print(models.summary())
+
+    # TensorBoard  earlytopping ModelCheckpoint
+
+    log_dir = "callbacks"
+    if not os.path.exists(log_dir):
+        os.mkdir(log_dir)
+    ouput_model_file = os.path.join(log_dir, "fashion_mnist.h5")
+
+    callbacks = [
+        keras.callbacks.TensorBoard(log_dir),
+        keras.callbacks.ModelCheckpoint(ouput_model_file, save_best_only=True),
+        keras.callbacks.EarlyStopping(patience=5, min_delta=1e-3),
+    ]
+    history = models.fit(x_train_scaled, y_train, epochs=20, validation_data=(x_valid_scaled, y_valid),
+                         callbacks=callbacks)
+
+    # 测试集上评估
+    print(models.evaluate(x_test_scaled, y_test))
+
+    # 存储训练过程中的一些值  如损失 accury等
+    print(history.history)
+
+    pd.DataFrame(history.history).plot(figsize=(8, 5))
+    plt.grid(True)
+    plt.gca().set_ylim(0, 1)
+    plt.show()
+
+
 if __name__ == '__main__':
-    show_single_image(x_train[0])
-    # show_images(1,1,x_train,y_train,class_names)
+    train()
+    # show_single_image(x_train[0])
+    # show_images(10,5,x_train,y_train,class_names)