R-DFCIL: Relation-Guided Representation Learning for Data-Free Class Incremental Learning

Qiankun Gao, Chen Zhao, Bernard Ghanem, Jian Zhang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

14 Scopus citations

Abstract

Class-Incremental Learning (CIL) struggles with catastrophic forgetting when learning new knowledge, and Data-Free CIL (DFCIL) is even more challenging without access to the training data of previously learned classes. Though recent DFCIL works introduce techniques such as model inversion to synthesize data for previous classes, they fail to overcome forgetting due to the severe domain gap between the synthetic and real data. To address this issue, this paper proposes relation-guided representation learning (RRL) for DFCIL, dubbed R-DFCIL. In RRL, we introduce relational knowledge distillation to flexibly transfer the structural relation of new data from the old model to the current model. Our RRL-boosted DFCIL can guide the current model to learn representations of new classes better compatible with representations of previous classes, which greatly reduces forgetting while improving plasticity. To avoid the mutual interference between representation and classifier learning, we employ local rather than global classification loss during RRL. After RRL, the classification head is refined with global class-balanced classification loss to address the data imbalance issue as well as learn the decision boundaries between new and previous classes. Extensive experiments on CIFAR100, Tiny-ImageNet200, and ImageNet100 demonstrate that our R-DFCIL significantly surpasses previous approaches and achieves a new state-of-the-art performance for DFCIL.
Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science
PublisherSpringer Nature Switzerland
Pages423-439
Number of pages17
ISBN (Print)9783031200496
DOIs
StatePublished - Oct 28 2022

Bibliographical note

KAUST Repository Item: Exported on 2022-12-06
Acknowledged KAUST grant number(s): OSR-CRG2021-4648
Acknowledgements: This work was supported by the King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research (OSR) under Award No. OSR-CRG2021-4648 and the Shenzhen General Research Project (JCYJ20190808182805919).

Fingerprint

Dive into the research topics of 'R-DFCIL: Relation-Guided Representation Learning for Data-Free Class Incremental Learning'. Together they form a unique fingerprint.

Cite this