如何找到CNN模型（keras）的ROC曲线和AUC分数

Question

我是深度学习新手。我正在尝试为以下代码生成 ROC 曲线。我正在使用喀拉斯。 class大小为10，图像为大小为100*100*3的RGB图像。

我经历了[这个link][1]。我的问题也一样，但我找不到真正的标签。我是这个领域的新手，所以请帮助我。我还查看了[This for true label][2]。

我的程序的代码片段是：

target_size=(100,100,3)

train_generator = train_datagen.flow_from_directory('path',
    target_size=target_size[:-1],
    batch_size=16,
    class_mode='categorical',
    subset='training',
    seed=random_seed)

valid_generator = ...

test_generator = ...
n_classes = len(set(train_generator.classes))



input_layer = keras.layers.Input(shape=target_size)

conv2d_1 = keras.layers.Conv2D(filters=64, kernel_size=(3,3), strides=1, padding='same', 
activation='relu',
                       kernel_initializer='he_normal')(input_layer)

batchnorm_1 = keras.layers.BatchNormalization()(conv2d_1)
maxpool1=keras.layers.MaxPool2D(pool_size=(2,2))(batchnorm_1)


conv2d_2 = keras.layers.Conv2D(filters=32, kernel_size=(3,3), strides=1, padding='same', 
activation='relu',
                       kernel_initializer='he_normal')(maxpool1)
batchnorm_2 = keras.layers.BatchNormalization()(conv2d_2)

maxpool2=keras.layers.MaxPool2D(pool_size=(2,2))(batchnorm_2)


flatten = keras.layers.Flatten()(maxpool2)
dense_1 = keras.layers.Dense(256, activation='relu')(flatten)

dense_2 = keras.layers.Dense(n_classes, activation='softmax')(dense_1)



model = keras.models.Model(input_layer, dense_3)

model.compile(optimizer=keras.optimizers.Adam(0.001),
      loss='categorical_crossentropy',
      metrics=['acc'])
model.summary()

model.fit_generator(generator=train_generator, validation_data=valid_generator,
            epochs=200)
            
score = model.evaluate_generator(test_generator)

print(score)

现在请帮助我获取 AUC 分数和 ROC 曲线。 [1]: [2]:

Answer 1

score函数默认不提供roc和auc分数，需要单独计算。您可以查看以下代码片段来计算 roc 和 auc 分数并绘制其中的值。

from sklearn.metrics import roc_curve
y_pred_keras = model.predict(X_test).ravel()
fpr_keras, tpr_keras, thresholds_keras = roc_curve(y_test, y_pred_keras)

from sklearn.metrics import auc
auc_keras = auc(fpr_keras, tpr_keras)

import matplotlib.pyplot as plt
plt.figure(1)
plt.plot([0, 1], [0, 1], 'k--')
plt.plot(fpr_keras, tpr_keras, label='Keras (area = {:.3f})'.format(auc_keras))
plt.plot(fpr_rf, tpr_rf, label='RF (area = {:.3f})'.format(auc_rf))
plt.xlabel('False positive rate')
plt.ylabel('True positive rate')
plt.title('ROC curve')
plt.legend(loc='best')
plt.show()

Answer 2

添加此代码。希望有用。

import numpy as np
from sklearn import metrics

x, y = test_generator.next()
prediction = model.predict(x)

predict_label1 = np.argmax(prediction, axis=-1)
true_label1 = np.argmax(y, axis=-1)

y = np.array(true_label1)

scores = np.array(predict_label1)
fpr, tpr, thresholds = metrics.roc_curve(y, scores, pos_label=9)
roc_auc = metrics.auc(fpr, tpr)


plt.figure()
lw = 2
plt.plot(fpr, tpr, color='darkorange',
 lw=lw, label='ROC curve (area = %0.2f)' % roc_auc)
plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Receiver operating characteristic (ROC)')
plt.legend(loc="lower right")
plt.show()

如何找到CNN模型（keras）的ROC曲线和AUC分数

How to find the ROC curve and AUC score of CNN model (keras)

python

keras

conv-neural-network

roc

auc