Metraipolis › archive › dc_3550fd74cd2d
Latent-Aligned Reasoning for Multimodal Recommendation — arXiv:2609.04645
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.04645
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T22:35:57.889Z
- SHA-256 of the archived text
1681e298d9d4c6fd832cea604344e0c1f8eab5e3644c70f641e6d5f049bc1c42- Stable URL of this page
https://metraipolis.com/documento/dc_3550fd74cd2d
How to cite
Latent-Aligned Reasoning for Multimodal Recommendation — arXiv:2609.04645 — arXiv (en), https://arxiv.org/abs/2609.04645 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3550fd74cd2d (Metraipolis archive dc_3550fd74cd2d, sha256 1681e298d9d4c6fd832cea604344e0c1f8eab5e3644c70f641e6d5f049bc1c42)
BibTeX
@misc{dc3550fd74cd2d,
title = {Latent-Aligned Reasoning for Multimodal Recommendation — arXiv:2609.04645},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.04645},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3550fd74cd2d; sha256 1681e298d9d4c6fd832cea604344e0c1f8eab5e3644c70f641e6d5f049bc1c42},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.04645. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multimodal Vision-Language Models (VLMs) have demonstrated remarkable capabilities in cross-modal understanding, yet a fundamental challenge persists when applying them to recommendation: as representations propagate through multi-step reasoning, both visual and textual signals progressively attenuate - a phenomenon we term cross-modal dilution. To address this, we propose LARK (Latent-Aligned Reasoning frameworK), a two-stage latent reasoning framework with complementary alignment mechanisms within a single VLM. In the first stage, learnable latent tokens are interleaved with multi-step chain-of-thought (CoT) reasoning and explicitly aligned with a frozen vision encoder, serving as visual checkpoints that preserve perceptual details throughout the reasoning chain. In the second stage, the latent representations are projected via a bridge MLP and trained with item-to-item contrastive learning; to prevent the reasoning semantics from fading, intermediate features are aligned with the CoT hidden states from the first stage, anchoring the final embeddings to the model's own reasoning output. Experiments on three public benchmarks and one industrial dataset show that LARK achieves state-of-the-art performance across multiple recommendation architectures, with controlled ablations confirming the distinct contribution of each component. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.CL). Categorie arXiv: Information Retrieval (cs.IR) ; Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.04645 — 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 2026-10-16T22:21:16.320Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).