Saved in:
Bibliographic Details
Main Authors: Huang, Hejin, Zhang, Jusheng, Cai, Kaitong, Wang, Jian, Pan, Rong
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.29259
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914433694957568
author Huang, Hejin
Zhang, Jusheng
Cai, Kaitong
Wang, Jian
Pan, Rong
author_facet Huang, Hejin
Zhang, Jusheng
Cai, Kaitong
Wang, Jian
Pan, Rong
contents Preference-based alignment objectives have been widely adopted, from RLHF-style pairwise learning in large language models to emerging applications in recommender systems. Yet, existing work rarely examines how Direct Preference Optimization (DPO) behaves under implicit feedback, where unobserved items are not reliable negatives. We conduct systematic experiments on multimodal sequential recommendation to compare common negative-selection strategies and their interaction with DPO training. Our central finding is that a simple modification, replacing deterministic hard negatives with stochastic sampling from a dynamic top-K candidate pool, consistently improves ranking performance. We attribute its effectiveness to two factors: (1) reducing erroneous suppressive gradients caused by false negatives, and (2) retaining informative hard signals while smoothing optimization via controlled stochasticity. With an optional sparse Mixture-of-Experts encoder for efficient capacity scaling, RoDPO achieves up to 5.25% NDCG@5 on three Amazon benchmarks, with nearly unchanged inference cost.
format Preprint
id arxiv_https___arxiv_org_abs_2603_29259
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Aligning Multimodal Sequential Recommendations via Robust Direct Preference Optimization with Sparse MoE
Huang, Hejin
Zhang, Jusheng
Cai, Kaitong
Wang, Jian
Pan, Rong
Information Retrieval
Computation and Language
Preference-based alignment objectives have been widely adopted, from RLHF-style pairwise learning in large language models to emerging applications in recommender systems. Yet, existing work rarely examines how Direct Preference Optimization (DPO) behaves under implicit feedback, where unobserved items are not reliable negatives. We conduct systematic experiments on multimodal sequential recommendation to compare common negative-selection strategies and their interaction with DPO training. Our central finding is that a simple modification, replacing deterministic hard negatives with stochastic sampling from a dynamic top-K candidate pool, consistently improves ranking performance. We attribute its effectiveness to two factors: (1) reducing erroneous suppressive gradients caused by false negatives, and (2) retaining informative hard signals while smoothing optimization via controlled stochasticity. With an optional sparse Mixture-of-Experts encoder for efficient capacity scaling, RoDPO achieves up to 5.25% NDCG@5 on three Amazon benchmarks, with nearly unchanged inference cost.
title Aligning Multimodal Sequential Recommendations via Robust Direct Preference Optimization with Sparse MoE
topic Information Retrieval
Computation and Language
url https://arxiv.org/abs/2603.29259