python - 为什么 keras 模型编译但 fit_generator 命令抛出 'model not compiled runtime error' ？

Question

我已经在带有 tensorflow 后端的 keras 中设置了一个带有 K 折交叉验证的自定义 CNN。model.compile() 函数在开始训练之前被调用，但是 model.fit_generator() 函数调用导致运行时错误:“You must compile your model before using it.”

我使用 ImageDataGenerator 进行数据扩充，并使用 fit_generator 函数进行训练。

到目前为止，我发现的唯一相关问题与 tensorflow 有关 eager execution Keras 中似乎未启用的功能。

代码如下:

模型定义:

model = Sequential()
model.add(Conv2D(24, (5, 5), 
          strides=(1, 1), 
          padding="valid", 
          data_format="channels_last",
          activation='relu', 
          use_bias=True,
          ))#out=96
model.add(Dropout(.25))
model.add(MaxPooling2D(pool_size=(2, 2)))#out=48
model.add(Conv2D(32, (3, 3), 
          strides=(1, 1), 
          padding="valid", 
          data_format="channels_last",
          activation='relu', 
          use_bias=True,
          ))#out=46
model.add(MaxPooling2D(pool_size=(2, 2)))#out=23
model.add(Conv2D(48, (3, 3), 
          strides=(1, 1), 
          padding="valid", 
          data_format="channels_last",
          activation='relu', 
          use_bias=True,
          ))#out=21
model.add(MaxPooling2D(pool_size=(2, 2)))#padding???
model.add(Flatten())
model.add(Dense(3, activation='softmax'))

...这里将是数据初始化...

设置图像生成器:

datagen_training = ImageDataGenerator(
            rotation_range = 20,
            width_shift_range = 0.3,
            height_shift_range=0.3,
            zoom_range=0.2,
            fill_mode = "constant",
            cval = 0,
            vertical_flip = True,
            validation_split = 0.2
            )
datagen_training.fit(data)

模型设置和训练:

rmsprop = optimizers.RMSprop(lr=0.001)#docu says to only tune the learning rate
kf = KFold(n_splits=FOLDS, shuffle = True, random_state=78945)
model.compile(rmsprop, loss = losses.categorical_crossentropy, metrics=[metrics.categorical_accuracy])
acc_hist = []
while True:
    history = object()
    for train_idx, val_idx in kf.split(data, labels):
        x_train, y_train = data[train_idx], labels[train_idx]
        x_val, y_val = data[val_idx], labels[val_idx]
        data_iterator = datagen_training.flow(x_train, y_train, batch_size=BATCH_SIZE)
        history = model.fit_generator(data_iterator, steps_per_epoch=len(x_train) // BATCH_SIZE, epochs=1)
    acc_hist.append(history.history['categorical_accuracy'][0])
    #stop if accuracy doesn't change within 3 epochs
    if stopping_criterion_met:
        break

Answer 1

模型未编译是因为它未被构建，它未被构建是因为 input_shape 未在第一层中指定。我不确定你的数据的输入形状，但你的第一层这样的东西会让模型编译:

model = Sequential()
model.add(Conv2D(24, (5, 5), 
      strides=(1, 1), 
      padding="valid", 
      data_format="channels_last",
      activation='relu', 
      use_bias=True,
      input_shape=(100,100,1)
      ))#out=96