Self-Cure-Network
Self-Cure-Network copied to clipboard
三个问题,望不吝赐教,感谢!
question 1:论文里说“Mislabeled samples are marked in red solid rectangles and ambiguous samples in green dash ones”,请问怎么判定mislabeled还是ambiguous?
question 2:通过RR Loss来确保high-importance group 的均值至少比low-importance group的均值大δ1,这样子限制的目的和作用是什么?
question 3: 论文中提到的re-weighting是怎么实现的?是针对那些修改了标签的图片在之后的训练中re-weight吗?figure2上有画出re-weighting但论文中没有明确解释,所以我不太清楚很好奇。
问题有点多,希望作者能详细解答一下,谢谢!
这几天搞完几篇cvpr2022的final version回答你这些问题哈。多谢,可以给我来一份私聊电子邮件
CNing715 @.***>于2022年3月1日 周二上午11:36写道:
question 1:论文里说“Mislabeled samples are marked in red solid rectangles and ambiguous samples in green dash ones”,请问怎么判定mislabeled还是ambiguous?
question 2:通过RR Loss来确保high-importance group 的均值至少比low-importance group的均值大δ1,这样子限制的目的和作用是什么?
question 3: 论文中提到的re-weighting是怎么实现的?是针对那些修改了标签的图片在之后的训练中re-weight吗?figure2上有画出re-weighting但论文中没有明确解释,所以我不太清楚很好奇。
问题有点多,希望作者能详细解答一下,谢谢!
— Reply to this email directly, view it on GitHub https://github.com/kaiwang960112/Self-Cure-Network/issues/67, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4FDCET5AW6UGUXKF7LU5WGDZANCNFSM5PS3LDSA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
You are receiving this because you are subscribed to this thread.Message ID: @.***>
您好,对于问题2我也有同样的疑问,还有一个不太理解的地方就是为什么只在训练集进行标签重新标记操作,而在测试集按原来的标签进行测试,这样不会冲突吗?如果方便的话希望您能再回答一下,非常期待您的回复