Metraipolis › archive › dc_975a320a69e7
DSPrompt: Dynamic Soft Prompt Defense Against M-RAG Corruption — arXiv:2608.16536
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.16536
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T19:01:31.852Z
- SHA-256 of the archived text
35cf411139cf8cbac80212e49bd00b9cd81c90a513b3f51c90d6b094d50d49ac- Stable URL of this page
https://metraipolis.com/documento/dc_975a320a69e7
How to cite
DSPrompt: Dynamic Soft Prompt Defense Against M-RAG Corruption — arXiv:2608.16536 — arXiv (en), https://arxiv.org/abs/2608.16536 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_975a320a69e7 (Metraipolis archive dc_975a320a69e7, sha256 35cf411139cf8cbac80212e49bd00b9cd81c90a513b3f51c90d6b094d50d49ac)
BibTeX
@misc{dc975a320a69e7,
title = {DSPrompt: Dynamic Soft Prompt Defense Against M-RAG Corruption — arXiv:2608.16536},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.16536},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_975a320a69e7; sha256 35cf411139cf8cbac80212e49bd00b9cd81c90a513b3f51c90d6b094d50d49ac},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.16536. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multimodal Retrieval Augmented Generation (M-RAG) is increasingly vulnerable to adversarial attacks where malicious data are crafted to produce embeddings that align with benign entries in the vector space, deceiving retrieval and inducing harmful outputs. Existing defenses primarily operate at query time, relying on auxiliary detectors, similarity re-ranking, or feature-consistency checks. However, these approaches suffer from non-trivial inference overhead, generalize poorly to unseen attack strategies, and often assume specific attack distributions. To address this, we propose DSPrompt, a Dynamic Soft Prompt defense framework that directly reshapes the retriever's embedding semantics, without modifying the retrieval pipeline. It inserts few learnable soft prompts into each layer of the visual and textual encoders of a frozen retriever, utilizing a shallow-to-deep length schedule that is adaptive to the capacity in the model layers. These prompts are trained under a dynamic min-max scheme: an online multimodal attacker continually crafts hard adversarial documents against the current retriever, while the defender is updated to push such documents out of the top-k while preserving the ranking and diversity of benign evidence. Because the defended encoder can be pre-computed and indexed exactly as in standard dense retrieval, DSPrompt incurs no additional per-query optimization and introduces fewer than 1% additional parameters. Extensive experiments across four benchmarks and three representative poisoning attacks show that DSPrompt substantially reduces the attack success rate and poison retrieval rate while maintaining near-lossless retrieval utility and generation fidelity, consistently outperforming existing defense baselines at a fraction of their computational cost. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.CL). Categorie arXiv: Cryptography and Security (cs.CR) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2608.16536 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-18T18:06:57.935Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).