GAP-Prompt：用于高效持续学习的门控自适应提示

中文摘要 AI 辅助

持续学习面临灾难性遗忘这一长期挑战，即顺序任务更新会损害先前习得的知识。尽管结合预训练模型的基于提示的方法通过冻结主干网络提供了一种颇具吸引力的解决方案，但它们通常依赖静态的、任务级的提示策略，忽略了任务内部的细粒度多样性。本文提出了门控自适应提示（Gated Adaptive Prompting，GAP-Prompt），这是一种在提示过程中引入实例级适应性的新方法。GAP-Prompt由三个协同模块组成：（1）实例条件门控，为每张单独的图像动态确定最优的提示注入层；（2）动态知识融合，对当前和历史提示执行实例感知的聚合，实现跨任务的知识整合；（3）共享提示蒸馏，将基础知识锚定在早期共享层以减轻遗忘。在CIFAR-100、ImageNet-R和CUB-200基准上的广泛评估表明，GAP-Prompt始终达到了最先进的性能。值得注意的是，在细粒度的CUB-200数据集上，GAP-Prompt达到了87.29%的准确率，接近联合训练的上限（88.00%），并大幅优于现有方法。

英文摘要

Continual learning faces the persistent challenge of catastrophic forgetting, where sequential task updates degrade previously acquired knowledge. While prompt-based methods integrated with pre-trained models offer a compelling solution by freezing the backbone, they often rely on static, task-level prompting strategies that overlook fine-grained intra-task diversity. In this paper, we propose Gated Adaptive Prompting (GAP-Prompt), a novel method that introduces instance-level adaptability to the prompting process. GAP-Prompt consists of three synergistic modules: (1) instance-conditioned gating, which dynamically determines optimal prompt injection layers for each individual image; (2) dynamic knowledge fusion, which performs instance-aware aggregation of current and historical prompts, enabling knowledge integration across tasks; and (3) shared prompt distillation, which anchors foundational knowledge in early shared layers to mitigate forgetting. Extensive evaluations on CIFAR-100, ImageNet-R, and CUB-200 benchmarks demonstrate that GAP-Prompt consistently achieves state-of-the-art performance. Notably, on the fine-grained CUB-200 dataset, GAP-Prompt reaches 87.29% accuracy, approaching the joint training upper bound (88.00%) and outperforming existing methods by a significant margin.

发表机构

• Université d’Orléans（奥尔良大学）
• INSA CVL（中央卢瓦尔河谷国立应用科学学院）
• CY Cergy Paris University（塞尔吉-蓬图瓦兹大学）
• ENSEA（法国高等电子与数字学院）
• CNRS（法国国家科学研究中心）
• Université de Technologie de Troyes（特鲁瓦技术大学）

机构由 AI 辅助整理，请以论文原文为准。