The exemplar-free class incremental learning requires classification models to learn new class knowledge incrementally without retaining any old samples. Recently, the framework based on parallel one-class classifiers (POC), which trains a one-class classifier (OCC) independently for each category, has attracted extensive attention, since it can naturally avoid catastrophic forgetting. POC, however, suffers from weak discriminability and comparability due to its independent training strategy for different OOCs. To meet this challenge, we propose a new framework, named Discriminative and Comparable One-class classifiers for Incremental Learning (DisCOIL). DisCOIL follows the basic principle of POC, but it adopts variational auto-encoders (VAE) instead of other well-established one-class classifiers (e.g. deep SVDD), because a trained VAE can not only identify the probability of an input sample belonging to a class but also generate pseudo samples of the class to assist in learning new tasks. With this advantage, DisCOIL trains a new-class VAE in contrast with the old-class VAEs, which forces the new-class VAE to reconstruct better for new-class samples but worse for the old-class pseudo samples, thus enhancing the comparability. Furthermore, DisCOIL introduces a hinge reconstruction loss to ensure the discriminability. We evaluate our method extensively on MNIST, CIFAR10, and Tiny-ImageNet. The experimental results show that DisCOIL achieves state-of-the-art performance.


    Access

    Download


    Export, share and cite



    Title :

    Exemplar-free Class Incremental Learning via Discriminative and Comparable One-class Classifiers


    Contributors:
    Sun, Wenju (author) / Li, Qingyong (author) / Zhang, Jing (author) / Wang, Danyu (author) / Wang, Wen (author) / Geng, Yangli-ao (author)

    Publication date :

    2022


    Remarks:

    [J]. Pattern Recognition, 2023, 140: 109561


    Type of media :

    Preprint


    Type of material :

    Electronic Resource


    Language :

    English







    Discriminative Models for Multi-Class Object Layout

    Desai, C. / Ramanan, D. / Fowlkes, C. C. | British Library Online Contents | 2011