{"ID":2870391,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.12990","arxiv_id":"2509.12990","title":"Dual-Stage Reweighted MoE for Long-Tailed Egocentric Mistake Detection","abstract":"In this report, we address the problem of determining whether a user performs an action incorrectly from egocentric video data. To handle the challenges posed by subtle and infrequent mistakes, we propose a Dual-Stage Reweighted Mixture-of-Experts (DR-MoE) framework. In the first stage, features are extracted using a frozen ViViT model and a LoRA-tuned ViViT model, which are combined through a feature-level expert module. In the second stage, three classifiers are trained with different objectives: reweighted cross-entropy to mitigate class imbalance, AUC loss to improve ranking under skewed distributions, and label-aware loss with sharpness-aware minimization to enhance calibration and generalization. Their predictions are fused using a classification-level expert module. The proposed method achieves strong performance, particularly in identifying rare and ambiguous mistake instances. The code is available at https://github.com/boyuh/DR-MoE.","short_abstract":"In this report, we address the problem of determining whether a user performs an action incorrectly from egocentric video data. To handle the challenges posed by subtle and infrequent mistakes, we propose a Dual-Stage Reweighted Mixture-of-Experts (DR-MoE) framework. In the first stage, features are extracted using a f...","url_abs":"https://arxiv.org/abs/2509.12990","url_pdf":"https://arxiv.org/pdf/2509.12990v2","authors":"[\"Boyu Han\",\"Qianqian Xu\",\"Shilong Bao\",\"Zhiyong Yang\",\"Sicong Li\",\"Qingming Huang\"]","published":"2025-09-16T12:00:42Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.AI\",\"cs.LG\"]","methods":"[\"LoRA\"]","has_code":false,"code_links":[{"ID":609760,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2870391,"paper_url":"https://arxiv.org/abs/2509.12990","paper_title":"Dual-Stage Reweighted MoE for Long-Tailed Egocentric Mistake Detection","repo_url":"https://github.com/boyuh/DR-MoE","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}