.png){kind=icon}
返回
Knowledge aggregation networks for class incremental learning
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2023.109310.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Most existing class incremental learning methods rely on storing old exemplars to avoid catastrophic forgetting. However, these methods inevitably face the gradient conflict problem, the inherent conflict between new streaming knowledge and existing knowledge in the gradient direction. To alleviate gradient conflict, this paper reuses the previous knowledge and expands the branch to accommodate new concepts instead of fine-tuning the original models. Specifically, this paper designs a novel dual-branch network called Knowledge Aggregation Networks. The previously trained model is frozen as a branch to retain existing knowledge, and a consistent trainable network is constructed as the other branch to learn new concepts. An adaptive feature fusion module is adopted to dynamically balance the two branches' information during training. Moreover, a model compression stage maintains the dual-branch structure. Extensive experiments on CIFAR-10 0, ImageNet-Sub, and ImageNet show that our method significantly outperforms the other methods and effectively balances stability and plasticity. & COPY; 2023 Elsevier Ltd. All rights reserved.
Keyword:
Class incremental learning
Catastrophic forgetting
Dual-branch network
Knowledge aggregation
Model compression
期刊
IF:
7.6
论文数:
1.3W
被引数:
4.5W
机构
暂无机构信息
引用论文
FoCL: Feature-oriented continual learning for generative modelsFoCL: 生成模型的面向特征的持续学习.png)
PATTERN RECOGNITION
OAAI
PATTERN RECOGNITIONIF7.6
OAAI分享收藏A comprehensive study of class incremental learning algorithms for visual tasksNEURAL NETWORKSOAAI
NEURAL NETWORKSIF6.3
OAAI分享收藏LoGID: An adaptive framework combining local and global incremental learning for dynamic selection of ensembles of HMMsPATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏分享收藏
查看更多内容