Jul 2026· International Journal of Computer Vision· Vol 134· 0 citations· 107 references
Computer Science
TL;DR
This paper systematically identifies one of the fundamental challenges behind CIL, named feature collision, where the features learned by the current task-specific model may collide with those of the previous models, leading to forgetting of previously learned tasks and hindering the learning of new tasks.
Miles decouples the learnable modules with the pre-trained model, exploiting prior information from intermediate features of the backbone network to enable more flexible parameter expansion, and orchestrating an efficient expansion of the parameter space through guided optimization.
Kai Jiang, Zisong Lin, Hongyuan Zhang et al.· IEEE Transactions on Image P...· 0 citations
Class-Incremental Learning (CIL) aims to enable models to sequentially learn new tasks while retaining knowledge from previous ones. Recently, merging-based pre-trained CIL methods have gained significant attention due to their competitive performance and high inference efficiency. However, most existing approaches dec...
Si-Yu Zhang, Wen Wang, Wen-Ju Sun et al.· Proceedings of the Thirty-Fi...· 0 citations
As a paradigm in continual learning, class incremental learning (CIL) aims to assimilate tasks with mutually exclusive label spaces in sequence while preserving previously established knowledge. Mitigating forgetting in CIL fundamentally relies on transferring knowledge across tasks. A straightforward exemplar-based ap...
Fan-Kang Xu, Lu Jin, Yanpeng Sun et al.· IEEE Transactions on Image P...· 0 citations
FACET proposes an efficient replay-free task-conditioned feature consistency loss, aiming to mitigate catastrophic forgetting of the learned mixture distribution in the adapter's feature space, and demonstrates robust scalability.
Meta-learning has emerged as an effective solver for cross-domain few-shot learning (CD-FSL) tasks. Despite achieving obvious progress recently, the typical episodic learning paradigm often causes the feature embedding model collapsing into the simplicity bias pitfall, viz., the model tends to prioritize some shortcut...
Fei Zhou, Peng Wang, Lei Zhang et al.· IEEE Transactions on Image P...· 0 citations
The proposed CARL framework employs two parallel ARL branches and aggregates their Mahalanobis distances during inference to improve prediction performance and is compared with recent methods using three widely recognized datasets.
