keras 可视化模型结果，f1_score，recall，acc,acc_valid,checkpoint

冷不防 2022-05-17 07:50 107阅读 0赞

''' Created on 2018年8月8日 '''
    
    import keras
    from keras.models import Sequential
    from keras.layers import Dense
    import numpy as np
    import tflearn
    import tflearn.datasets.mnist as mnist
    from keras.callbacks import Callback,ModelCheckpoint
    from sklearn.metrics import f1_score, precision_score, recall_score
    import matplotlib.pyplot as plt
    
    class Metrics(Callback):
        def on_train_begin(self, logs={}):
            self.val_f1s = []
            self.val_recalls = []
            self.val_precisions = []
    
        def on_epoch_end(self, epoch, logs={}):
            val_predict = (np.asarray(self.model.predict(self.validation_data[0]))).round()
            val_targ = self.validation_data[1]
            _val_f1 = f1_score(val_targ, val_predict)
            _val_recall = recall_score(val_targ, val_predict)
            _val_precision = precision_score(val_targ, val_predict)
            self.val_f1s.append(_val_f1)
            self.val_recalls.append(_val_recall)
            self.val_precisions.append(_val_precision)
            print('- val_f1: %.4f - val_precision: %.4f - val_recall: %.4f'%(_val_f1, _val_precision, _val_recall))
            return
    
    #只拿0,1数据
    x_train, y_train, x_test, y_test = mnist.load_data(one_hot=False)
    train_index0 = np.where(y_train == 0)[0]
    train_index1 = np.where(y_train == 1)[0]
    test_index0 = np.where(y_test == 0)[0]
    test_index1 = np.where(y_test == 1)[0]
    print(len(train_index0))
    print(len(train_index1))
    train_indexs = np.append(train_index0,train_index1)
    test_indexs = np.append(test_index0,test_index1)
    print(len(train_indexs))
    
    x_train = x_train[train_indexs,:]
    y_train = y_train[train_indexs]
    x_test = x_test[test_indexs,:]
    y_test = y_test[test_indexs]
    model = Sequential()
    model.add(Dense(units=64, activation='relu', input_dim=784))
    model.add(Dense(units=1, activation='sigmoid'))
    model.compile(loss='binary_crossentropy',
                  optimizer='sgd',
                  metrics=['accuracy'])
    
    
    checkpoint_filepath = 'models/model-ep{epoch:03d}-loss{loss:.3f}-acc{acc:.3f}-val_loss{val_loss:.3f}-val_acc{val_acc:.3f}.h5'
    checkpoint = ModelCheckpoint(checkpoint_filepath, monitor='acc', verbose=1, save_best_only=True, mode='max')
    
    metrics = Metrics()
    
    history = model.fit(x_train, y_train,
                 epochs=10,
                 batch_size=32,
                 validation_data=(x_test, y_test),
                 callbacks=[metrics,checkpoint])
    # print(history.history.keys())
    plt.plot(history.history['acc'],'b--')
    plt.plot(history.history['val_acc'],'y-')
    plt.plot(metrics.val_f1s,'r.-')
    plt.plot(metrics.val_precisions,'g-')
    plt.plot(metrics.val_recalls,'c-')
    
    plt.title('DenseNet201 model report')
    plt.ylabel('evaluation')
    plt.xlabel('epoch')
    plt.legend(['train_accuracy', 'val_accuracy','val_f1-score','val_precisions','val_recalls'], loc='lower right')
    plt.savefig('results/result_acc.png')
    plt.show()