{"ID":2853393,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.16446","arxiv_id":"2510.16446","title":"VIPAMIN: Visual Prompt Initialization via Embedding Selection and Subspace Expansion","abstract":"In the era of large-scale foundation models, fully fine-tuning pretrained networks for each downstream task is often prohibitively resource-intensive. Prompt tuning offers a lightweight alternative by introducing tunable prompts while keeping the backbone frozen. However, existing visual prompt tuning methods often fail to specialize the prompts or enrich the representation space--especially when applied to self-supervised backbones. We show that these limitations become especially pronounced in challenging tasks and data-scarce settings, where effective adaptation is most critical. In this work, we introduce VIPAMIN, a visual prompt initialization strategy that enhances adaptation of self-supervised models by (1) aligning prompts with semantically informative regions in the embedding space, and (2) injecting novel representational directions beyond the pretrained subspace. Despite its simplicity--requiring only a single forward pass and lightweight operations--VIPAMIN consistently improves performance across diverse tasks and dataset sizes, setting a new state of the art in visual prompt tuning. Our code is available at https://github.com/iamjaekyun/vipamin.","short_abstract":"In the era of large-scale foundation models, fully fine-tuning pretrained networks for each downstream task is often prohibitively resource-intensive. Prompt tuning offers a lightweight alternative by introducing tunable prompts while keeping the backbone frozen. However, existing visual prompt tuning methods often fai...","url_abs":"https://arxiv.org/abs/2510.16446","url_pdf":"https://arxiv.org/pdf/2510.16446v1","authors":"[\"Jaekyun Park\",\"Hye Won Chung\"]","published":"2025-10-18T10:49:48Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.LG\"]","methods":"[]","has_code":false,"code_links":[{"ID":608079,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2853393,"paper_url":"https://arxiv.org/abs/2510.16446","paper_title":"VIPAMIN: Visual Prompt Initialization via Embedding Selection and Subspace Expansion","repo_url":"https://github.com/iamjaekyun/vipamin","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}