{"ID":2884565,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.06036","arxiv_id":"2508.06036","title":"More Is Better: A MoE-Based Emotion Recognition Framework with Human Preference Alignment","abstract":"In this paper, we present our solution for the semi-supervised learning track (MER-SEMI) in MER2025. We propose a comprehensive framework, grounded in the principle that \"more is better,\" to construct a robust Mixture of Experts (MoE) emotion recognition system. Our approach integrates a diverse range of input modalities as independent experts, including novel signals such as knowledge from large Vision-Language Models (VLMs) and temporal Action Unit (AU) information. To effectively utilize unlabeled data, we introduce a consensus-based pseudo-labeling strategy, generating high-quality labels from the agreement between a baseline model and Gemini, which are then used in a two-stage training paradigm. Finally, we employ a multi-expert voting ensemble combined with a rule-based re-ranking process to correct prediction bias and better align the outputs with human preferences. Evaluated on the MER2025-SEMI challenge dataset, our method achieves an F1-score of 0.8772 on the test set, ranking 2nd in the track. Our code is available at https://github.com/zhuyjan/MER2025-MRAC25.","short_abstract":"In this paper, we present our solution for the semi-supervised learning track (MER-SEMI) in MER2025. We propose a comprehensive framework, grounded in the principle that \"more is better,\" to construct a robust Mixture of Experts (MoE) emotion recognition system. Our approach integrates a diverse range of input modaliti...","url_abs":"https://arxiv.org/abs/2508.06036","url_pdf":"https://arxiv.org/pdf/2508.06036v1","authors":"[\"Jun Xie\",\"Yingjian Zhu\",\"Feng Chen\",\"Zhenghao Zhang\",\"Xiaohui Fan\",\"Hongzhu Yi\",\"Xinming Wang\",\"Chen Yu\",\"Yue Bi\",\"Zhaoran Zhao\",\"Xiongjun Guan\",\"Zhepeng Wang\"]","published":"2025-08-08T05:44:26Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Mixture of Experts\",\"Language Model\"]","has_code":false,"code_links":[{"ID":611090,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2884565,"paper_url":"https://arxiv.org/abs/2508.06036","paper_title":"More Is Better: A MoE-Based Emotion Recognition Framework with Human Preference Alignment","repo_url":"https://github.com/zhuyjan/MER2025-MRAC25","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}