Cus_acc = 1-keras.backend.mean (keras.backend.abs((y_pred − y_true)/0.50), axis = −1).The reason might be overfitting of outliers in the training data.
確定！ 回上一頁
