CHEEM：复用、新建、适配与跳过的持续学习——一种分层探索-利用方法
CHEEM: Continual Learning by Reuse, New, Adapt and Skip - A Hierarchical Exploration-Exploitation Approach
把神经架构搜索当作持续学习中的内部记忆：每个新任务通过 reuse/new/adapt/skip 四种原语动态改写骨干网络结构，视觉研究者可将其作为免样本类增量学习下替代 prompt 微调的骨干自适应方案，并复用其任务质心外部记忆做任务 ID 推理。
原文依据 a HEE-guided efficient neural architecture search (HEE-NAS) that enables a learning-to-adapt backbone via four primitive operations - reuse, new, adapt, and skip
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论为有效应对真实动态环境的复杂性，持续学习必须从性质各异的任务流中增量地获取、更新并积累知识，同时不遭受对先验知识的灾难性遗忘。这一能力对人类认知而言是与生俱来的，但对现代深度学习系统仍是一项重大挑战。该挑战的核心在于稳定性-可塑性困境：需要在利用先验知识、整合新信息，以及依据任务复杂度与协同性自适应分配模型容量之间取得平衡。本文提出一种新的无样本类增量持续学习（ExfCCL）框架，通过分层探索-利用（HEE）方法解决上述问题。方法核心是由 HEE 引导的高效神经架构搜索（HEE-NAS），它借助复用、新建、适配与跳过四种原语，使骨干网络具备学习如何自适应的能力，从而充当内部记忆，在流式任务间动态更新所选组件。为解决 ExfCCL 中的任务 ID 推理问题，我们利用现有工作中提出的任务质心外部记忆。我们将方法命名为 CHEEM（持续分层探索-利用记忆）。CHEEM 在具有挑战性的 MTIL 与 VDD 基准上，使用 Tiny 与 Base 两种视觉 Transformer 以及所提出的整体品质因数（FoM）指标进行评估。它显著优于最优的基于提示的持续学习方法，并接近全量微调的上界。此外，它以语义上有意义的方式学习到针对各个任务定制的自适应模型结构。
英文原文摘要 1732 字符
To effectively manage the complexities of real-world dynamic environments, continual learning must incrementally acquire, update, and accumulate knowledge from a stream of tasks of different nature - without suffering from catastrophic forgetting of prior knowledge. While this capability is innate to human cognition, it remains a significant challenge for modern deep learning systems. At the heart of this challenge lies the stability-plasticity dilemma: the need to balance leveraging prior knowledge, integrating novel information, and allocating model capacity adaptively based on task complexity and synergy. In this paper, we propose a novel exemplar-free class-incremental continual learning (ExfCCL) framework that addresses these issues through a Hierarchical Exploration-Exploitation (HEE) approach. The core of our method is a HEE-guided efficient neural architecture search (HEE-NAS) that enables a learning-to-adapt backbone via four primitive operations - reuse, new, adapt, and skip - thereby serving as an internal memory that dynamically updates selected components across streaming tasks. To address the task ID inference problem in ExfCCL, we exploit an external memory of task centroids proposed in the prior art. We term our method CHEEM (Continual Hierarchical-Exploration-Exploitation Memory). CHEEM is evaluated on the challenging MTIL and VDD benchmarks using both Tiny and Base Vision Transformers and a proposed holistic Figure-of-Merit (FoM) metric. It significantly outperforms state-of-the-art prompting-based continual learning methods, closely approaching full fine-tuning upper bounds. Furthermore, it learns adaptive model structures tailored to individual tasks in a semantically meaningful way.