BPG：面向领域增量学习的可塑性与泛化能力平衡框架
BPG: Balancing Plasticity and Generalization for Domain Incremental Learning
- College of Artificial Intelligence, Xi’an Jiaotong University（西安交通大学人工智能学院）
- Shenzhen University of Advanced Technology（深圳先进技术大学）
- Harbin Institute of Technology, Shenzhen（哈尔滨工业大学（深圳））
- School of Software Engineering, Xi’an Jiaotong University（西安交通大学软件工程学院）
机构由 AI 辅助整理，请以论文原文为准。
中文总结 AI 辅助
本研究提出BPG框架，通过动态适配的BPG-Adapter与软领域混合的BPG-Inference解决领域增量学习的参数冗余或能力不足问题，在多数据集上实现最优平均准确率并大幅降低遗忘率。
中文摘要 AI 辅助
深度神经网络在各类任务中表现优异，但难以在不断变化的数据分布间实现泛化，在领域偏移下会出现显著的性能下降。领域增量学习（Domain Incremental Learning, DIL）旨在让模型在持续适配新领域的同时保留已有知识，以此应对这一挑战。现有DIL方法中，参数隔离范式达到了最优性能，但这类方法常采用一刀切的策略适配新领域，导致要么学习能力不足，要么参数冗余。本研究提出BPG这一统一框架，通过两个互补组件解决上述问题：BPG-Adapter根据各领域的特征可分性动态确定对应适配器的隐藏层维度；BPG-Inference是一种软领域混合策略，在测试时整合多个领域特定模型，缓解领域ID误选问题。在DomainNet、CDDB和CORe50数据集上的实验结果表明，BPG的性能始终优于基于统一适配器的方法和硬领域选择策略，达到了最优的平均准确率，同时在DomainNet上将遗忘率降至低至0.22%。
英文摘要
Deep neural networks excel in various tasks but struggle to generalize across evolving data distributions, leading to significant performance degradation under domain shifts. Domain incremental learning (DIL) addresses this challenge by enabling models to continuously adapt while retaining prior knowledge. Among existing DIL approaches, the parameter-isolation paradigm achieves state-of-the-art performance. However, these methods often adopt a one-size-fits-all approach to adapt to new domains, resulting in either insufficient learning capacity or redundant parameters. In this work, we propose BPG, a unified framework that addresses both challenges through two complementary components: BPG-Adapter, which dynamically determines each domain's adapter hidden dimension based on domain-specific feature separability, and BPG-Inference, a soft domain mixture strategy that integrates multiple domain-specific models at test time, mitigating domain ID misselection. Experimental results on DomainNet, CDDB, and CORe50 demonstrate that BPG consistently outperforms uniform adapter-based approaches and hard domain selection strategies, achieving state-of-the-art average accuracy while reducing forgetting to as low as 0.22% on DomainNet.
