Hands-On Generative Adversarial Networks with Keras
上QQ阅读APP看书,第一时间看更新

Categorical crossentropy loss

The categorical crossentropy loss function measures the weighted divergence between targets, , and predictions, , where denotes the data point, and denotes the class. For single label classification problems, the targets operate as an indicator variable and the loss is reduced to :