Bilateral Memory Consolidation for Continual Learning
Xing Nie1,2; Shixiong Xu1,2; Xiyan Liu3; Gaofeng Meng1,2,4; Chunlei Huo1,2; Shiming Xiang1,2
2023-06
Conference NameIEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR)
Conference Date2023年6月18日–2023年6月22日
Conference PlaceMontreal, Canada
Abstract

Humans are proficient at continuously acquiring and integrating new knowledge.  By contrast, deep models forget catastrophically, especially when tackling highly long task sequences.  Inspired by the way our brains constantly rewrite and consolidate past recollections, we propose a novel Bilateral Memory Consolidation (BiMeCo) framework that focuses on enhancing memory interaction capabilities.  Specifically, BiMeCo explicitly decouples model parameters into short-term memory module and long-term memory module, responsible for representation ability of the model and generalization over all learned tasks, respectively.  BiMeCo encourages dynamic interactions between two memory modules by knowledge distillation and momentum-based updating for forming generic knowledge to prevent forgetting.  The proposed BiMeCo is parameterefficient and can be integrated into existing methods seamlessly.  Extensive experiments on challenging benchmarks show that BiMeCo significantly improves the performance ofexisting continual learning methods.  For example, combined with the state-of-the-art method CwD [55], BiMeCo brings in significant gains ofaround 2% to 6% while using 2x fewer parameters on CIFAR-100 under ResNet-18.

Language英语
Sub direction classification图像视频处理与分析
planning direction of the national heavy laboratory智能进化环境
Paper associated data
Document Type会议论文
Identifierhttp://ir.ia.ac.cn/handle/173211/57462
Collection多模态人工智能系统全国重点实验室_先进时空数据分析与学习
Affiliation1.State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences.
2.School of Artificial Intelligence, University of Chinese Academy of Sciences.
3.Baidu Inc., China.
4.Centre for Artificial Intelligence and Robotics, HK Institute of Science & Innovation, CAS.
First Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Recommended Citation
GB/T 7714
Xing Nie,Shixiong Xu,Xiyan Liu,et al. Bilateral Memory Consolidation for Continual Learning[C],2023.
Files in This Item:
File Name/Size DocType Version Access License
Bilateral Memory Con(1031KB)会议论文 开放获取CC BY-NC-SAView
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Xing Nie]'s Articles
[Shixiong Xu]'s Articles
[Xiyan Liu]'s Articles
Baidu academic
Similar articles in Baidu academic
[Xing Nie]'s Articles
[Shixiong Xu]'s Articles
[Xiyan Liu]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Xing Nie]'s Articles
[Shixiong Xu]'s Articles
[Xiyan Liu]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: Bilateral Memory Consolidation for Continual Learning.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.