Sculpting Margin Penalty: Intra-Task Adapter Merging and Classifier Calibration for Few-Shot Class-Incremental Learning

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Bai, Liang, Song, Hong, Li, Jinfu, Lin, Yucong, Fan, Jingfan, Fu, Tianyu, Ai, Danni, Xiao, Deqiang, Yang, Jian
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913978831077376
author Bai, Liang
Song, Hong
Li, Jinfu
Lin, Yucong
Fan, Jingfan
Fu, Tianyu
Ai, Danni
Xiao, Deqiang
Yang, Jian
author_facet Bai, Liang
Song, Hong
Li, Jinfu
Lin, Yucong
Fan, Jingfan
Fu, Tianyu
Ai, Danni
Xiao, Deqiang
Yang, Jian
contents Real-world applications often face data privacy constraints and high acquisition costs, making the assumption of sufficient training data in incremental tasks unrealistic and leading to significant performance degradation in class-incremental learning. Forward-compatible learning, which prospectively prepares for future tasks during base task training, has emerged as a promising solution for Few-Shot Class-Incremental Learning (FSCIL). However, existing methods still struggle to balance base-class discriminability and new-class generalization. Moreover, limited access to original data during incremental tasks often results in ambiguous inter-class decision boundaries. To address these challenges, we propose SMP (Sculpting Margin Penalty), a novel FSCIL method that strategically integrates margin penalties at different stages within the parameter-efficient fine-tuning paradigm. Specifically, we introduce the Margin-aware Intra-task Adapter Merging (MIAM) mechanism for base task learning. MIAM trains two sets of low-rank adapters with distinct classification losses: one with a margin penalty to enhance base-class discriminability, and the other without margin constraints to promote generalization to future new classes. These adapters are then adaptively merged to improve forward compatibility. For incremental tasks, we propose a Margin Penalty-based Classifier Calibration (MPCC) strategy to refine decision boundaries by fine-tuning classifiers on all seen classes' embeddings with a margin penalty. Extensive experiments on CIFAR100, ImageNet-R, and CUB200 demonstrate that SMP achieves state-of-the-art performance in FSCIL while maintaining a better balance between base and new classes.
format Preprint
id arxiv_https___arxiv_org_abs_2508_05094
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Sculpting Margin Penalty: Intra-Task Adapter Merging and Classifier Calibration for Few-Shot Class-Incremental Learning
Bai, Liang
Song, Hong
Li, Jinfu
Lin, Yucong
Fan, Jingfan
Fu, Tianyu
Ai, Danni
Xiao, Deqiang
Yang, Jian
Computer Vision and Pattern Recognition
Real-world applications often face data privacy constraints and high acquisition costs, making the assumption of sufficient training data in incremental tasks unrealistic and leading to significant performance degradation in class-incremental learning. Forward-compatible learning, which prospectively prepares for future tasks during base task training, has emerged as a promising solution for Few-Shot Class-Incremental Learning (FSCIL). However, existing methods still struggle to balance base-class discriminability and new-class generalization. Moreover, limited access to original data during incremental tasks often results in ambiguous inter-class decision boundaries. To address these challenges, we propose SMP (Sculpting Margin Penalty), a novel FSCIL method that strategically integrates margin penalties at different stages within the parameter-efficient fine-tuning paradigm. Specifically, we introduce the Margin-aware Intra-task Adapter Merging (MIAM) mechanism for base task learning. MIAM trains two sets of low-rank adapters with distinct classification losses: one with a margin penalty to enhance base-class discriminability, and the other without margin constraints to promote generalization to future new classes. These adapters are then adaptively merged to improve forward compatibility. For incremental tasks, we propose a Margin Penalty-based Classifier Calibration (MPCC) strategy to refine decision boundaries by fine-tuning classifiers on all seen classes' embeddings with a margin penalty. Extensive experiments on CIFAR100, ImageNet-R, and CUB200 demonstrate that SMP achieves state-of-the-art performance in FSCIL while maintaining a better balance between base and new classes.
title Sculpting Margin Penalty: Intra-Task Adapter Merging and Classifier Calibration for Few-Shot Class-Incremental Learning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.05094