VALSE 论文速览 第135期：基于自异构混合专家网络的长尾视觉识别 ...

论文题目：Long-Tailed Visual Recognition via Self-Heterogeneous Integration with Knowledge Excavation

作者列表：

金焱 (厦门大学)，李梦柯 (深圳光明实验室)，卢杨 (厦门大学)，张晓明 (香港浸会大学)，王菡子 (厦门大学)

B站观看网址：

https://www.bilibili.com/video/BV15G411m7uf/

论文摘要：

Deep neural networks have made huge progress in the last few decades. However, as the real-world data often exhibits a long-tailed distribution, vanilla deep models tend to be heavily biased toward the majority classes. To address this problem, state-of-the-art methods usually adopt a mixture of experts (MoE) to focus on different parts of the long-tailed distribution. Experts in these methods are with the same model depth, which neglects the fact that different classes may have different preferences to be fit by models with different depths. To this end, we propose a novel MoE-based method called Self-Heterogeneous Integration with Knowledge Excavation (SHIKE). We first propose Depth-wise Knowledge Fusion (DKF) to fuse features between different shallow parts and the deep part in one network for each expert, which makes experts more diverse in terms of representation. Based on DKF, we further propose Dynamic Knowledge Transfer (DKT) to reduce the influence of the hardest negative class that has a non-negligible impact on the tail classes in our MoE framework. As a result, the classification accuracy of long-tailed data can be significantly improved, especially for the tail classes. SHIKE achieves the state-of-the-art performance of 56.3%, 60.3%, 75.4%, and 41.9% on CIFAR100-LT (IF100), ImageNet-LT, iNaturalist 2018, and Places-LT, respectively. 

论文信息：

[1] Yan Jin, Mengke Li, Yang Lu, Yiu-ming Cheung, and Hanzi Wang, “Long-Tailed Visual Recognition via Self-Heterogeneous Integration with Knowledge Excavation,” in Proceedings of IEEE/ CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, Canada, June 18-22, 2023.

论文链接：

[https://arxiv.org/abs/2304.01279]

代码链接：

[https://github.com/jinyan-06/SHIKE]

视频讲者简介：

金焱，厦门大学信息学院硕士生。主要研究方向为深度长尾学习。

特别鸣谢本次论文速览主要组织者：

月度轮值AC：叶茫 (武汉大学)