Publication Type

Conference Proceeding Article

Version

acceptedVersion

Publication Date

6-2021

Abstract

Class-Incremental Learning (CIL) aims to learn a classification model with the number of classes increasing phase-by-phase. An inherent problem in CIL is the stability-plasticity dilemma between the learning of old and new classes, i.e., high-plasticity models easily forget old classes, but high-stability models are weak to learn new classes. We alleviate this issue by proposing a novel network architecture called Adaptive Aggregation Networks (AANets) in which we explicitly build two types of residual blocks at each residual level (taking ResNet as the baseline architecture): a stable block and a plastic block. We aggregate the output feature maps from these two blocks and then feed the results to the next-level blocks. We adapt the aggregation weights in order to balance these two types of blocks, i.e., between stability and plasticity, dynamically. We conduct extensive experiments on three CIL benchmarks: CIFAR-100, ImageNet-Subset, and ImageNet, and show that many existing CIL methods can be straightforwardly incorporated into the architecture of AANets to boost their performances.

Keywords

Adaptation models, Computer vision, Adaptive systems, Computer architecture, Network architecture, Benchmark testing, Stability analysis

Discipline

Artificial Intelligence and Robotics | OS and Networks

Research Areas

Intelligent Systems and Optimization

Publication

2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition: Virtual, June 21-24: Proceedings

First Page

2544

Last Page

2553

ISBN

9781665445092

Identifier

10.1109/CVPR46437.2021.00257

Publisher

IEEE

City or Country

Piscataway, NJ

Copyright Owner and License

Authors

Additional URL

https://doi.org/10.1109/CVPR46437.2021.00257

Download
Download Research Data

Share

COinS