Non-Forgetting Knowledge Allocation with Bi-level Competition for Class-Incremental Learning 事件

Non-Forgetting Knowledge Allocation with Bi-level Competition for Class-Incremental Learning arXiv:2605.29592v1 Announce Type: new Abstract: Class-Incremental Learning (CIL) with pre-trained models (PTMs) aims to sequentially adapt PTMs to new categories without forgetting old knowledge. Built upon PTMs, existing adapter-based methods mainly train models via distinct task-specific adapters, and present a uniform knowledge allocation for each adapter during inference. However, this allocation me