30/11/2020

Class-incremental Learning with Rectified Feature-Graph Preservation

Cheng-Hsun Lei, Yi-Hsin Chen, Wen-Hsiao Peng, Wei-Chen Chiu

Keywords:

Abstract Paper Similar Papers

Abstract: In this paper, we address the problem of distillation-based class-incremental learning with a single head. A central theme of this task is to learn new classes that arrive in sequential phases over time while keeping the model's capability of recognizing seen classes with only limited memory for preserving seen data samples. Many regularization strategies have been proposed to mitigate the phenomenon of catastrophic forgetting. To understand better the essence of these regularizations, we introduce a feature-graph preservation perspective. Insights into their merits and faults motivate our weighted-Euclidean regularization for old knowledge preservation. We further propose rectified cosine normalization and show how it can work with binary cross-entropy to increase class separation for effective learning of new classes. Experimental results on both CIFAR-100 and ImageNet datasets demonstrate that our method outperforms the state-of-the-art approaches in reducing classification error, easing catastrophic forgetting, and encouraging evenly balanced accuracy over different classes. Our project page is at : https://github.com/yhchen12101/FGP-ICL.

The video of this talk cannot be embedded. You can watch it here:
https://accv2020.github.io/miniconf/poster_38.html
(Link will open in new window)
 0
 0
 0
 0
  Share    
This is an embedded video. Talk and the respective paper are published at ACCV 2020 virtual conference. If you are one of the authors of the paper and want to manage your upload, see the question "My papertalk has been externally embedded..." in the FAQ section.

Comments

Post Comment
no comments yet
code of conduct: tbd Characters remaining: 140

Similar Papers

 0
 0
 0
 0
 8:32 
 0
 0
 0
 0
 4:23 
 0
 0
 0
 0
 7:35 
 0
 0
 0
 0
 5:13 
 0
 0
 0
 0
 15:13 
 0
 0
 0
 0
 17:42 
 0
 0
 0
 0
 10:21 
 0
 0
 0
 0
 5:04 
 0
 0
 0
 0
 5:10 
 0
 0
 0
 1
 5:00 
 0
 0
 0
 0
 8:29 
 0
 0
 0
 0
 3:17 
 0
 0
 0
 0
 11:27 
 0
 0
 0
 0
 14:34 
 0
 0
 0
 0
 3:05 
 0
 0
 0
 0
 5:51 
 0
 0
 0
 0
 5:06 
 0
 0
 0
 0
 5:06 
 0
 0
 0
 0
 8:28 
 0
 0
 0
 0
 1:01 
 0
 0
 0
 0
 3:24 
 0
 0
 0
 0
 5:01 
 0
 0
 0
 0
 4:44 
 0
 0
 0
 0
 8:15 
 0
 0
 0
 0
 3:19 
 0
 0
 0
 0
 10:06 
 0
 0
 0
 0
 15:59 
 0
 0
 0
 0
 3:06 
 0
 0
 0
 0
 1:01 
 0
 0
 0
 0
 15:34 
 0
 0
 0
 0
 6:02 
 0
 0
 0
 0
 2:56 
 0
 0
 0
 0
 10:53 
 0
 0
 0
 1
 5:13 
 0
 0
 0
 0
 2:54 
 0
 0
 0
 0
 13:42 
 0
 0
 0
 0
 17:12 
 1
 1
 1
 1
 5:20 
 0
 0
 0
 0
 0:59 
 0
 0
 0
 0
 13:59