Domain-Incremental Learning

Overview

Domain-incremental learning studies how a model can keep performing the same task as its input distribution changes, such as across locations, devices, environments, or time. Common approaches replay selected examples, constrain or distill updates to protect prior knowledge, or isolate new domain-specific parameters with adapters, prompts, or expert models. Other methods use prototypes or routing to select domain-specific knowledge, while memory-free approaches adapt frozen foundation models or preserve feature structure. The central challenge is balancing adaptation to new domains against forgetting old ones, especially when domain boundaries are unclear and storage, computation, or access to past data is limited.

Latest papers

61 papers5 papers in the last four weeks, down 17% on the four weeks before. 0.0% of all new papers.

May 29, 2024

AdalReID: Auto-selected Knowledge Adapters for Lifelong Person Re-identification

Lifelong Person Re-Identification (LReID) extends traditional ReID by requiring systems to continually learn from non-overlapping datasets across different times and locations, adapting to new identities while preserving knowledge of previous ones. Existing approaches, either rehearsal-free or rehearsal-based, still suffer from the problem of catastrophic forgetting since they try to cram diverse knowledge into one fixed model. To overcome this limitation, we introduce a novel framework AdalReID, that adopts knowledge adapters and a parameter-free auto-selection mechanism for lifelong learning. Concretely, we incrementally build distinct adapters to learn domain-specific knowledge at each step, which can effectively learn and preserve knowledge across different datasets. Meanwhile, the proposed auto-selection strategy adaptively calculates the knowledge similarity between the input set and the adapters. On the one hand, the appropriate adapters are selected for the inputs to process ReID, and on the other hand, the knowledge interaction and fusion between adapters are enhanced to improve the generalization ability of the model. Extensive experiments are conducted to demonstrate the superiority of our AdalReID, which significantly beats SOTAs by about 5% and 3% mAP on seen and unseen domains.