Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning

About

Real-world applications require the classification model to adapt to new classes without forgetting old ones. Correspondingly, Class-Incremental Learning (CIL) aims to train a model with limited memory size to meet this requirement. Typical CIL methods tend to save representative exemplars from former classes to resist forgetting, while recent works find that storing models from history can substantially boost the performance. However, the stored models are not counted into the memory budget, which implicitly results in unfair comparisons. We find that when counting the model size into the total budget and comparing methods with aligned memory size, saving models do not consistently work, especially for the case with limited memory budgets. As a result, we need to holistically evaluate different CIL methods at different memory scales and simultaneously consider accuracy and memory size for measurement. On the other hand, we dive deeply into the construction of the memory buffer for memory efficiency. By analyzing the effect of different layers in the network, we find that shallow and deep layers have different characteristics in CIL. Motivated by this, we propose a simple yet effective baseline, denoted as MEMO for Memory-efficient Expandable MOdel. MEMO extends specialized layers based on the shared generalized representations, efficiently extracting diverse representations with modest cost and maintaining representative exemplars. Extensive experiments on benchmark datasets validate MEMO's competitive performance. Code is available at: https://github.com/wangkiw/ICLR23-MEMO

Da-Wei Zhou, Qi-Wei Wang, Han-Jia Ye, De-Chuan Zhan• 2022

Related benchmarks

TaskDatasetResultRank
Class-incremental learningCIFAR-100
Averaged Incremental Accuracy75.83
234
Class-incremental learningImageNet-100
Avg Acc81.11
74
Class-incremental learningCUB
Avg Accuracy81.4
45
Class-incremental learningImageNet-100 B=50, C=10 1.0
Avg Incremental Acc76.83
42
Class-incremental learningCIFAR-100 B0_Inc5
Average Accuracy67.42
36
Class-incremental learningCIFAR
Accuracy69.8
28
Class-incremental learningCIFAR-100 B50Inc10
Accuracy (t=5)0.5812
24
Domain-incremental learningCORe50
Avg Accuracy (A)64.8
22
Online Continual LearningCIFAR-10
Average AUC73.21
20
Online Continual LearningCIFAR-100
AAUC40.6
20
Showing 10 of 26 rows

Other info

Follow for update