keras中没有对AUC的定义,需要我们自己定义:
x = np.linspace(-6, 6, 200)
y = np.array([0.0]*100 + [1.0]*100)
state = np.random.get_state()
np.random.shuffle(x)
np.random.set_state(state)
np.random.shuffle(y)
x_train, y_train = x[0:160], y[0:160]
plt.scatter(x_train, y_train)
plt.show()
from keras.models import Sequential
from keras.layers import Dense
import tensorflow as tf
from sklearn.metrics import roc_auc_score
def auroc(y_true, y_pred):
return tf.compat.v1.py_func(roc_auc_score, (y_true, y_pred), tf.double)
model = Sequential()
model.add(Dense(units=1, input_dim=1))
from keras.optimizers import SGD
model.compile(loss=my_loss, optimizer=SGD(lr=0.01, momentum=0.9, nesterov=True),metrics=['accuracy', auroc])
model.fit(x_train, y_train, epochs=100, batch_size=64)
# model.train_on_batch(x_batch, y_batch)
‘py_func’的问题,因为tf1升到tf2,需要把tf.py_func改成tf.compat.v1.pyfunc即可。
参考文档:
https://cloud.tencent.com/developer/ask/154717
https://blog.csdn.net/huazaikai/article/details/106162826