Few-Shot Class-Incremental Learning with Prior Knowledge

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Jiang, Wenhao, Li, Duo, Hu, Menghan, Zhai, Guangtao, Yang, Xiaokang, Zhang, Xiao-Ping
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909090831138816
author Jiang, Wenhao
Li, Duo
Hu, Menghan
Zhai, Guangtao
Yang, Xiaokang
Zhang, Xiao-Ping
author_facet Jiang, Wenhao
Li, Duo
Hu, Menghan
Zhai, Guangtao
Yang, Xiaokang
Zhang, Xiao-Ping
contents To tackle the issues of catastrophic forgetting and overfitting in few-shot class-incremental learning (FSCIL), previous work has primarily concentrated on preserving the memory of old knowledge during the incremental phase. The role of pre-trained model in shaping the effectiveness of incremental learning is frequently underestimated in these studies. Therefore, to enhance the generalization ability of the pre-trained model, we propose Learning with Prior Knowledge (LwPK) by introducing nearly free prior knowledge from a few unlabeled data of subsequent incremental classes. We cluster unlabeled incremental class samples to produce pseudo-labels, then jointly train these with labeled base class samples, effectively allocating embedding space for both old and new class data. Experimental results indicate that LwPK effectively enhances the model resilience against catastrophic forgetting, with theoretical analysis based on empirical risk minimization and class distance measurement corroborating its operational principles. The source code of LwPK is publicly available at: \url{https://github.com/StevenJ308/LwPK}.
format Preprint
id arxiv_https___arxiv_org_abs_2402_01201
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Few-Shot Class-Incremental Learning with Prior Knowledge
Jiang, Wenhao
Li, Duo
Hu, Menghan
Zhai, Guangtao
Yang, Xiaokang
Zhang, Xiao-Ping
Machine Learning
Artificial Intelligence
To tackle the issues of catastrophic forgetting and overfitting in few-shot class-incremental learning (FSCIL), previous work has primarily concentrated on preserving the memory of old knowledge during the incremental phase. The role of pre-trained model in shaping the effectiveness of incremental learning is frequently underestimated in these studies. Therefore, to enhance the generalization ability of the pre-trained model, we propose Learning with Prior Knowledge (LwPK) by introducing nearly free prior knowledge from a few unlabeled data of subsequent incremental classes. We cluster unlabeled incremental class samples to produce pseudo-labels, then jointly train these with labeled base class samples, effectively allocating embedding space for both old and new class data. Experimental results indicate that LwPK effectively enhances the model resilience against catastrophic forgetting, with theoretical analysis based on empirical risk minimization and class distance measurement corroborating its operational principles. The source code of LwPK is publicly available at: \url{https://github.com/StevenJ308/LwPK}.
title Few-Shot Class-Incremental Learning with Prior Knowledge
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2402.01201