自定义指标:在tf.keras中使用scikit的AucRoc计算器

import tensorflow as tf
from sklearn.metrics import roc_auc_score

class AUC_ROC(tf.keras.metrics.Metric):
    def __init__(self, name='auc_roc', **kwargs):
        super(AUC_ROC, self).__init__(name=name, **kwargs)
        self.true_positives = self.add_weight(name='tp', initializer='zeros')
        self.false_positives = self.add_weight(name='fp', initializer='zeros')
        self.true_negatives = self.add_weight(name='tn', initializer='zeros')
        self.false_negatives = self.add_weight(name='fn', initializer='zeros')

    def update_state(self, y_true, y_pred, sample_weight=None):
        y_true = tf.cast(y_true, tf.float32)
        y_pred = tf.cast(y_pred, tf.float32)

        self.true_positives.assign_add(tf.reduce_sum(tf.cast(tf.logical_and(tf.equal(y_true, 1), tf.equal(tf.round(y_pred), 1)), tf.float32)))
        self.false_positives.assign_add(tf.reduce_sum(tf.cast(tf.logical_and(tf.equal(y_true, 0), tf.equal(tf.round(y_pred), 1)), tf.float32)))
        self.true_negatives.assign_add(tf.reduce_sum(tf.cast(tf.logical_and(tf.equal(y_true, 0), tf.equal(tf.round(y_pred), 0)), tf.float32)))
        self.false_negatives.assign_add(tf.reduce_sum(tf.cast(tf.logical_and(tf.equal(y_true, 1), tf.equal(tf.round(y_pred), 0)), tf.float32)))

    def result(self):
        auc_roc = roc_auc_score(tf.cast(tf.round(self.true_positives + self.false_negatives), tf.int32),
                                tf.cast(tf.round(self.true_positives + self.false_positives), tf.int32))
        return auc_roc

    def reset_states(self):
        self.true_positives.assign(0)
        self.false_positives.assign(0)
        self.true_negatives.assign(0)
        self.false_negatives.assign(0)

model.compile(optimizer='adam',
              loss='binary_crossentropy',
              metrics=[AUC_ROC()])