很玄學，沒有修改參數，在test上的准確率從98%多變為99.1%了
參考鏈接：《簡單粗暴Tensorflow》，狂吹
import tensorflow as tf import numpy as np class MNISTLoader(): def __init__(self): mnist = tf.keras.datasets.mnist (self.train_data, self.train_label), (self.test_data, self.test_label) = mnist.load_data() # MNIST中的圖像默認為uint8（0-255的數字）。以下代碼將其歸一化到0-1之間的浮點數，並在最后增加一維作為顏色通道 self.train_data = np.expand_dims(self.train_data.astype(np.float32) / 255.0, axis=-1) # [60000, 28, 28, 1] self.test_data = np.expand_dims(self.test_data.astype(np.float32) / 255.0, axis=-1) # [10000, 28, 28, 1] self.train_label = self.train_label.astype(np.int32) # [60000] self.test_label = self.test_label.astype(np.int32) # [10000] self.num_train_data, self.num_test_data = self.train_data.shape[0], self.test_data.shape[0] def get_batch(self, batch_size): # 從數據集中隨機取出batch_size個元素並返回 index = np.random.randint(0, np.shape(self.train_data)[0], batch_size) return self.train_data[index, :], self.train_label[index] # tf.keras.layers.Conv2D( # filters, kernel_size, strides=(1, 1), padding='valid', data_format=None, # dilation_rate=(1, 1), activation=None, use_bias=True, # kernel_initializer='glorot_uniform', bias_initializer='zeros', # kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, # kernel_constraint=None, bias_constraint=None, **kwargs # ) # filters: Integer, the dimensionality of the output space (i.e. the number of output filters in the convolution). # kernel_size：感受野 An integer or tuple/list of 2 integers, specifying the height and width of the 2D convolution window. # strides：橫縱步幅 # padding：valid:表示不夠卷積核大小的塊,則丟棄;same表示不夠卷積核大小的塊就補0,所以輸出和輸入形狀相同 class CNN(tf.keras.Model): def __init__(self): super().__init__() self.conv1 = tf.keras.layers.Conv2D( filters=32, # 卷積層神經元（卷積核）數目 kernel_size=[5, 5], # 感受野大小 padding='same', # padding策略（vaild 或 same） activation=tf.nn.relu # 激活函數 ) self.pool = tf.keras.layers.MaxPool2D(pool_size=[2, 2], strides=2) self.conv2 = tf.keras.layers.Conv2D( filters=64, kernel_size=[5, 5], padding='same', activation=tf.nn.relu ) self.flatten = tf.keras.layers.Reshape(target_shape=(7 * 7 * 64,)) self.dense1 = tf.keras.layers.Dense(units=1024, activation=tf.nn.relu) self.dense2 = tf.keras.layers.Dense(units=10) def call(self, inputs): x = self.conv1(inputs) # [batch_size, 28, 28, 32] x = self.pool(x) # [batch_size, 14, 14, 32] x = self.conv2(x) # [batch_size, 14, 14, 64] x = self.pool(x) # [batch_size, 7, 7, 64] x = self.flatten(x) # [batch_size, 7 * 7 * 64] x = self.dense1(x) # [batch_size, 1024] x = self.dense2(x) # [batch_size, 10] output = tf.nn.softmax(x) return output num_epochs = 5 batch_size = 50 learning_rate = 0.001 model = CNN() data_loader = MNISTLoader() optimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate) num_batches = int(data_loader.num_train_data // batch_size * num_epochs) for batch_index in range(num_batches): X, y = data_loader.get_batch(batch_size) #從數據集中隨機選一部分數據 with tf.GradientTape() as tape: y_pred = model(X) #得到預測值 loss = tf.keras.losses.sparse_categorical_crossentropy(y_true=y, y_pred=y_pred)#計算loss loss = tf.reduce_mean(loss) print("batch %d: loss %f" % (batch_index, loss.numpy())) grads = tape.gradient(loss, model.variables) #calc grads optimizer.apply_gradients(grads_and_vars=zip(grads, model.variables)) #update grads #tf.keras.metrics.SparseCategoricalAccuracy是一個評估器 #不太懂為什么分批檢查預測值 sparse_categorical_accuracy = tf.keras.metrics.SparseCategoricalAccuracy() num_batches = int(data_loader.num_test_data // batch_size) for batch_index in range(num_batches): start_index, end_index = batch_index * batch_size, (batch_index + 1) * batch_size #model.predict 輸入測試數據,輸出預測結果 y_pred = model.predict(data_loader.test_data[start_index: end_index]) sparse_categorical_accuracy.update_state(y_true=data_loader.test_label[start_index: end_index], y_pred=y_pred) print("test accuracy: %f" % sparse_categorical_accuracy.result())