在訓練較大網路時, 往往想載入預訓練的模型, 但若想在網路結構上做些添補, 可能出現問題一二...

一下是添補的幾種情形, 此處以單輸出迴歸任務為例:

# 添在末尾:
base_model = InceptionV3(weights='imagenet', include_top=False)
x = base_model.output
x = GlobalAveragePooling2D()(x)
x = Dense(1, activation='relu')(x)

model = Model(inputs=base_model.input, outputs=x)
model.summary()
 

# 添在開頭和末尾:
# 在開頭加1x1卷積層, 使4通道降為3通道, 再傳入InceptionV3
def head_model(input_shape=(150, 150, 4)):
    input_tensor = Input(input_shape)
    x = Conv2D(128, (1, 1), activation='relu')(input_tensor)
    x = Conv2D(3, (1, 1), activation='relu')(x)
    model = Model(inputs=input_tensor, outputs=x, name='head')
    return model

head_model = head_model()
body_model = InceptionV3(weights='imagenet', include_top=False)
base_model = Model(head_model.input, body_model(head_model.output))
x = base_model.output
x = GlobalAveragePooling2D()(x)
x = Dense(1, activation='relu')(x)
model = Model(inputs=body_model.inputs, outputs=x)
model.summary()
 

# 兩資料輸入流合併於末尾:
base_model = InceptionV3(weights='imagenet', include_top=False, input_shape=(150, 150, 3))
flat = Flatten()(base_model.output)
input_K = Input((100, ))    # another_input
K_flow = Activation(activation='linear')(input_K)
x = concatenate([flat, K_flow])    # 合流
x = Dense(1024, activation='relu')(x)
x = Dense(512, activation='relu')(x)
x = Dense(1, activation='relu')(x)
model = Model(inputs=[*base_model.inputs, input_K], outputs=x)    # 資料生成器那裡也以這種形式生成([x_0, x_1], y)即可.
model.summary()
 

References:
末尾
開頭
末尾合流_0 末尾合流_1