arXiv：cs.AI· Minghao Kong, Jiurun Chen, Ying Gao, Xiangbin Meng, Rongjie Wang·· 5 小时前AI 评分34
低成本视频-时间先验成为熟悉视频 EEG-fNIRS 情绪回归的强基线
Low-Cost Video--Time Priors as a Strong Baseline for EEG--fNIRS Emotion Regression on Familiar Videos
AI 导读
一项研究提出以视频身份和视频内时间构成的低成本"视频-时间先验"作为熟悉视频情绪回归的强基线，在 24 名被试的五折留一评估中，其 MAE 与 EEG-fNIRS 融合结果分别相差 0.05 和 0.32 以内。消融实验显示，视频身份与视频内时间贡献了大部分性能提升，而 EEG-fNIRS 带来的增益较小且因被试和视频而异。研究据此将 EEG-fNIRS 定位为熟悉视频情绪回归的可选残差信号。
正文
Abstract:Continuous emotion regression estimates moment-to-moment valence and arousal while a viewer watches a video. In familiar-video deployment, responses fron training participant-specific estimate, and prior-dominating fixed fusion tests whether physiology adds residual correction. In five-fold subject-held-out evaluation on 24was within 0.05 and 0.32 MAE of fusion in the internal and external evaluations, respectively. Source-explicit ablations showed that video identity and within-video tine accounted for most of the reduction, while EG-FNIRS gains were smaller and varied across participants and videos. These results identify the video-time prior as a strong, low-cost baseline and position EEG-fNIRS as an optional residual signal for familiar-video emotion regression.
| Subjects: | Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM) |
| Cite as: | arXiv:2610.03618 [cs.AI] |
| (or arXiv:2610.03618v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.03618 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Minghao Kong [view email]
[v1]
Fri, 2 Oct 2026 17:14:45 UTC (1,490 KB)
来源：arXiv：cs.AI · arxiv.org