{"ID":6267032,"CreatedAt":"2026-07-10T01:11:38.759438437Z","UpdatedAt":"2026-07-13T01:02:08.706470581Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.08107","arxiv_id":"2607.08107","title":"BACH: A Bayesian Admixture of Contrastive Heads for Multi-Interest Two-Tower Retrieval","abstract":"Two-tower retrievers compress each user into a single embedding, limiting their ability to serve diverse interests. Multi-interest models give each user several heads scored by a maximum inner product, but their hard-routing training under-utilizes heads (routing collapse) and gives no per-user estimate of how much each interest matters for serving. We present \\textbf{BACH} (\\emph{Bayesian Admixture of Contrastive Heads}), which casts multi-interest two-tower retrieval as a per-user mixture over the heads, fit by variational inference. The soft mixture trains every head (mitigating collapse), produces a per-user weighting of the interests that is reused at serving, and admits a shared global-codebook variant with precomputable retrieval. On three large-scale benchmarks, MovieLens-20M, Taobao, and Netflix, BACH improves top-of-ranking retrieval over hard-routing multi-interest and single-vector baselines at every head count; we further find that scoring every candidate by its best head, consistent with serving, outperforms the usual target-routed training, and that BACH improves further still.","short_abstract":"Two-tower retrievers compress each user into a single embedding, limiting their ability to serve diverse interests. Multi-interest models give each user several heads scored by a maximum inner product, but their hard-routing training under-utilizes heads (routing collapse) and gives no per-user estimate of how much eac...","url_abs":"https://arxiv.org/abs/2607.08107","url_pdf":"https://arxiv.org/pdf/2607.08107v1","authors":"[\"Quoc Phong Nguyen\",\"Paul Albert\",\"Long Vuong\",\"Vuong Le\",\"Julien Monteil\"]","published":"2026-07-09T04:48:04Z","proceeding":"cs.IR","tasks":"[\"cs.IR\",\"cs.LG\"]","methods":"[]","has_code":false}