PulseAugur
EN
LIVE 09:48:00

New BACH model enhances multi-interest retrieval systems

By PulseAugur Editorial · [2 sources] ·

Researchers have introduced BACH, a novel Bayesian admixture model designed to improve multi-interest two-tower retrieval systems. Unlike existing methods that can suffer from routing collapse and underutilization of user interest heads, BACH employs a soft mixture approach trained via variational inference. This method ensures all heads are trained, provides per-user interest weightings for serving, and has demonstrated superior performance on large-scale benchmarks including MovieLens-20M, Taobao, and Netflix. AI

IMPACT Improves retrieval accuracy and efficiency for systems handling diverse user interests.

RANK_REASON The cluster describes a new academic paper detailing a novel retrieval model.

Read on arXiv cs.IR (Information Retrieval) →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New BACH model enhances multi-interest retrieval systems

COVERAGE [2]

  1. arXiv cs.LG TIER_1 English(EN) · Quoc Phong Nguyen, Paul Albert, Long Vuong, Vuong Le, Julien Monteil ·

    BACH: A Bayesian Admixture of Contrastive Heads for Multi-Interest Two-Tower Retrieval

    arXiv:2607.08107v1 Announce Type: cross 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 train…

  2. arXiv cs.IR (Information Retrieval) TIER_1 English(EN) · Julien Monteil ·

    BACH: A Bayesian Admixture of Contrastive Heads for Multi-Interest Two-Tower Retrieval

    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 gi…

RELATED ENTITIES

RELATED TOPICS