Class-Incremental Learning with Rectified Feature-Graph Preservation

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

*Corresponding author for this work

    Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


    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 :

    Original languageEnglish
    Title of host publicationComputer Vision – ACCV 2020 - 15th Asian Conference on Computer Vision, 2020, Revised Selected Papers
    EditorsHiroshi Ishikawa, Cheng-Lin Liu, Tomas Pajdla, Jianbo Shi
    PublisherSpringer Science and Business Media Deutschland GmbH
    Number of pages17
    ISBN (Print)9783030695439
    StatePublished - 2021
    Event15th Asian Conference on Computer Vision, ACCV 2020 - Virtual, Online
    Duration: 30 Nov 20204 Dec 2020

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume12627 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349


    Conference15th Asian Conference on Computer Vision, ACCV 2020
    CityVirtual, Online


    Dive into the research topics of 'Class-Incremental Learning with Rectified Feature-Graph Preservation'. Together they form a unique fingerprint.

    Cite this