arXiv 2605.295922026-05-28cs.CV

非遗忘知识分配与双层竞争用于类增量学习

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

South China University of Technology（华南理工大学）；Hong Kong Polytechnic University（香港理工大学）；Agency for Science, Technology and Research（科技研究局）；Microsoft（微软）；Guilin University of Electronic Technology（桂林电子科技大学）；Sun Yat-sen University（中山大学）

Xiang Tan, Run He, Yawen Cui, Mengchen Zhao, Yan Wu, Tianyi Chen, Huiping Zhuang, Xiaonan Luo, Guanbin Li

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对基于预训练模型的类增量学习中适配器知识分配不均和遗忘问题，提出非遗忘分配与双层竞争方法（NoFA-BC），通过递归最小二乘构建非遗忘分配器，并引入任务内赢家通吃和任务间最后淘汰机制优化适配器利用。

摘要：

基于预训练模型（PTM）的类增量学习（CIL）旨在顺序地将PTM适应到新类别而不遗忘旧知识。现有的基于适配器的方法主要通过不同的任务特定适配器训练模型，并在推理时为每个适配器呈现统一的知识分配。然而，这种分配机制忽略了任务差异的本质，导致适配器的利用次优。此外，在CIL约束下，分配器在任务演化时容易遗忘。为了解决这些问题，我们提出了一种具有双层竞争的非遗忘分配（NoFA-BC）。NoFA-BC通过将分配器训练转化为递归最小二乘问题来构建非遗忘分配器（NFA），并实现了与使用所有数据训练等效的分配器。基于NFA，提出了双层竞争（BLC），包括任务内级别的赢家通吃（WTA）机制和任务间级别的最后淘汰（LOF）消除，以提供更好的适配器知识分配。WTA提取任务内最显著的logit来表示适配器的贡献，LOF抑制不相关的适配器。通过BLC，每个适配器的参与比例可以根据每个输入进行调整。此外，还加入了稳定性增强（SE）过程，以进一步提高旧任务的性能。

英文摘要：

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 mechanism ignores the nature of task discrepancy and leads to suboptimal utilization of adapters. Also, under CIL constraint, an allocator is prone to forgetting when tasks evolve. To address these issues, we propose a Non-Forgetting Allocation with Bi-Level Competition (NoFA-BC). NoFA-BC constructs a non-forgetting allocator (NFA) by transforming the allocator training into a recursive least-squares problem and achieves an allocator equivalent to that trained with all data. Based on the NFA, a Bi-Level Competition (BLC) including an intra-task level Winner-Takes-All (WTA) mechanism and inter-task Last-Ones-Fall (LOF) elimination is proposed to provide better allocation of adapter knowledge. WTA extracts the most significant logit within a task to represent the adapter's contribution and LOF suppresses the irrelevant adapters. With BLC, participation ratio of each adapter can be tailored for each input. Moreover, a Stability Enhancement (SE) process is incorporated to further improve the performance of old tasks.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。