1、log loss
2、WBE loss
带权重的交叉熵
3、Focal loss
容易过拟合?不知why
4、DIce loss
对于肿块太小的梯度容易变化剧烈
5、IOU loss
6、Tversky loss
7、敏感性-特异性损失
8、Generalized Dice loss
9、BCE + Dice loss
10、Dice + Focal loss
11、Exponential Logarithmic loss
原文地址:https://www.cnblogs.com/double-t/p/10907756.html
时间: 2024-10-09 19:31:35