search

LetheSec / Fer2013-Facial-Emotion-Recognition-Pytorch

This method achieves SOTA single model accuracy of 73.70 % on FER2013 without using extra training data.
MIT License
84 stars 16 forks source link

mixup_data #2

Closed jiantenggei closed 2 years ago

jiantenggei commented 2 years ago

hi ! i want know the theory of mixup_data function in utils.py 你好 我想问一下 mixup_data 的原理是什么。

LetheSec commented 2 years ago

是用到了mixup的技术，来自这篇文章：mixup: Beyond Empirical Risk Minimization

jiantenggei commented 2 years ago

好的 我了解一下 十分感谢你的回复

------------------ Original ------------------ From: Lethe @.> Date: Mon,May 9,2022 3:56 PM To: LetheSec/Fer2013-Facial-Emotion-Recognition-Pytorch @.> Cc: Jian Tengfei @.>, Author @.> Subject: Re: [LetheSec/Fer2013-Facial-Emotion-Recognition-Pytorch] mixup_data(Issue #2)

是用到了mixup的技术，来自这篇文章：mixup: Beyond Empirical Risk Minimization

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>