We present the Tamed Cross Entropy (TCE) loss function, a robust derivative of the standard Cross Entropy (CE) loss used in deep learning for classification tasks. However, unlike other robust losses, the TCE loss is designed to exhibit the same training properties than the CE loss in ...
We present the Tamed Cross Entropy (TCE) loss function, a robust derivative of the standard Cross Entropy (CE) loss used in deep learning for classification tasks. However, unlike other robust losses, the TCE loss is designed to exhibit the same training properties than the CE loss in noisele...
LVQ(E,G,Z)=‖x−x^‖2+‖sg[E(x)]−zq‖22+β‖sg[zq]−E(x)‖22 其中Lrec=‖x−x^‖2为重建损失(reconstruction loss),而sg[·]为梯度终止操作(stop-gradient operation)。之所以要在E(x)和zq之间加入sg[·]操作,是因为我们在这两个特征间进行了离散化转换,如果直接进行...
We present the Tamed Cross Entropy (TCE) loss function, a robust derivative of the standard Cross Entropy (CE) loss used in deep learning for classification tasks. However, unlike other robust losses, the TCE loss is designed to exhibit the same training properties than the CE loss in ...