原来损失函数还可以这么玩
收藏
A General and Adaptive Robust Loss Function
这篇文章介绍的l1、l2 自适应loss函数能很好的降低异常值对拟合曲线的影响
0
收藏
请登录后评论
还有解读博客
https://mp.weixin.qq.com/s?__biz=MzI5MDUyMDIxNA==&mid=2247508223&idx=3&sn=cfd79a2879d78b0cf9892f9ea87b6f5e&chksm=ec1c4d06db6bc41056cb485d60ad3ead474fdf156bc2a0c1ddb11f4e9a60cadfc68e6cc56286&mpshare=1&scene=1&srcid=0921OMem8rZDZZiPw9VwyPbl&sharer_sharetime=1600651645350&sharer_shareid=6161cb162d6309a81e4971fcc370f099&exportkey=A7C9zQGKWhDI5kxMSd%2FZ%2BQU%3D&pass_ticket=7kcOXruy3ZUsqdRSqNhMfwr%2FQuhmUiHNNNqWC0ev2nHmn1LqB5zPnXnRzzAYPaWh&wx_header=0#rd
感谢分享
感觉自适应的方法越来越常用了