keras遇到的问题

1、保存keras模型并重新导入，发现自己定义的损失函数不能用

ValueError: Unknown loss function:bp_mll_loss

解决：加一个custom_objects

# model.save('BPMLL_add_noise_less_fea_locate10.h5')
# from keras.models import load_model
# model = load_model('BPMLL_add_noise_less_fea_locate10.h5')
model.save('BPMLL_add_noise_less_fea_locate10.h5')
from keras.models import load_model
model = load_model('BPMLL_add_noise_less_fea_locate10.h5',custom_objects={'bp_mll_loss': bp_mll_loss})

2、keras将最好的迭代的性能的模型保存

3 code

def CNN(X_train,y_train,X_val,y_val,X_test,y_test):
    print('***************************************************training************************************************')

    import keras
    from keras.datasets import mnist
    from keras.models import Sequential
    from keras.layers import Dense, Dropout, Flatten, Activation
    from keras.layers import Conv1D, MaxPooling1D
    from keras import backend as K
    from keras.layers import LeakyReLU
    from keras.layers.normalization import BatchNormalization

    batch_size = 128
    num_classes = 41
    epochs = 200
    input_shape = (1024,10)

    y_train = keras.utils.to_categorical(y_train, num_classes)
    y_val = keras.utils.to_categorical(y_val, num_classes)
    y_test = keras.utils.to_categorical(y_test, num_classes)

    model = Sequential()
    model.add(Conv1D(filters = 32, kernel_size=16,padding='same',input_shape=input_shape))
    model.add(LeakyReLU(alpha=0.01))
    model.add(Conv1D(filters = 32, kernel_size=16,padding='same'))
    model.add(BatchNormalization())
    model.add(LeakyReLU(alpha=0.01))
    model.add(MaxPooling1D(pool_size=4, strides=4, padding='valid'))

    model.add(Conv1D(filters = 64, kernel_size=16,padding='same'))
    model.add(LeakyReLU(alpha=0.01))
    model.add(Conv1D( filters = 64, kernel_size=16,padding='same'))
    model.add(BatchNormalization())
    model.add(LeakyReLU(alpha=0.01))
    model.add(MaxPooling1D(pool_size=4, strides=4, padding='valid'))

    model.add(Conv1D(filters = 128, kernel_size=16,padding='same'))
    model.add(LeakyReLU(alpha=0.01))
    model.add(Conv1D( filters = 128, kernel_size=16,padding='same'))
    model.add(BatchNormalization())
    model.add(LeakyReLU(alpha=0.01))
    model.add(MaxPooling1D(pool_size=4, strides=4, padding='valid'))         

    model.add(Flatten())
    model.add(Dense(256))
    model.add(LeakyReLU(alpha=0.01))
    model.add(Dense(128))
    model.add(LeakyReLU(alpha=0.01))
    model.add(Dense(num_classes, activation='softmax'))

    model.compile(loss=keras.losses.categorical_crossentropy,
              optimizer=keras.optimizers.Adam(lr=5*10**-6, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False),
              metrics=['accuracy'])

    model.summary()


   early_stopping = keras.callbacks.EarlyStopping(
    monitor='val_loss',
    min_delta=0.00001,
    patience=500, 
    verbose=0, 
    mode='auto'
    )

callbacks_list = [early_stopping]
    

    model.fit(X_train, y_train,
          batch_size=batch_size,
          callbacks = callbacks_list,
          epochs=epochs,
          verbose=1,
          validation_data=(X_val, y_val))

    score = model.evaluate(X_test, y_test, verbose=0)
    print('Test loss:', score[0])
    print('Test accuracy:', score[1])

    prediction = model.predict_classes(X_test, verbose=1)
    prediction = keras.utils.to_categorical(prediction, num_classes)
    
    from sklearn.metrics import accuracy_score
    print ('************Accuracy of CNN classifier:',accuracy_score(y_test,prediction))

    from sklearn.metrics import f1_score,recall_score,precision_score
    print('Pre_micro = ',precision_score(y_test, prediction, average='micro'))
    print('Rec_micro = ',recall_score(y_test, prediction, average='micro'))
    print('F1_micro = ',f1_score(y_test, prediction, average='micro'))

    from sklearn.metrics import classification_report
    print (classification_report(y_test,prediction))
    model.save('../model_40/cnn_add_noise.h5')