{"ID":2840002,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.14401","arxiv_id":"2511.14401","title":"Language as an Anchor: Preserving Relative Visual Geometry for Domain Incremental Learning","abstract":"A key challenge in Domain Incremental Learning (DIL) is to continually learn under shifting distributions while preserving knowledge from previous domains. Existing methods face a fundamental dilemma. On one hand, projecting all domains into a single unified visual space leads to inter-domain interference and semantic distortion, as large shifts may vary with not only visual appearance but also underlying semantics. On the other hand, isolating domain-specific parameters causes knowledge fragmentation, creating \"knowledge islands\" that hamper knowledge reuse and exacerbate forgetting. To address this issue, we propose LAVA (Language-Anchored Visual Alignment), a novel DIL framework that replaces direct feature alignment with relative alignment driven by a text-based reference anchor. LAVA guides the visual representations of each incoming domain to preserve a consistent relative geometry, which is defined by mirroring the pairwise semantic similarities between the class names. This anchored geometric structure acts as a bridge across domains, enabling the retrieval of class-aware prior knowledge and facilitating robust feature aggregation. Extensive experiments on standard DIL benchmarks demonstrate that LAVA achieves significant performance improvements over state-of-the-arts. Code is available at https://github.com/ShuyiGeng/LAVA.","short_abstract":"A key challenge in Domain Incremental Learning (DIL) is to continually learn under shifting distributions while preserving knowledge from previous domains. Existing methods face a fundamental dilemma. On one hand, projecting all domains into a single unified visual space leads to inter-domain interference and semantic...","url_abs":"https://arxiv.org/abs/2511.14401","url_pdf":"https://arxiv.org/pdf/2511.14401v1","authors":"[\"Shuyi Geng\",\"Tao Zhou\",\"Yi Zhou\"]","published":"2025-11-18T12:06:55Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[]","has_code":false,"code_links":[{"ID":606936,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2840002,"paper_url":"https://arxiv.org/abs/2511.14401","paper_title":"Language as an Anchor: Preserving Relative Visual Geometry for Domain Incremental Learning","repo_url":"https://github.com/ShuyiGeng/LAVA","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}