MER-R1: Multimodal Emotion Reasoning via Slow-Fast Thinking Synergy
Specifically, for reasoning-based MLLMs, fast thinking by triggering direct answers often outperforms slow thinking after deliberative reasoning. Our empirical analyses show that fast thinking improves recall with broader and more confident predictions, whereas slow thinking favors precision through conservative filtering of incorrect categories. Building on these insights, we propose MER-R1, a reinforcement learning framework that turns slow-fast complementarity into explicit optimization.
- ▪Specifically, for reasoning-based MLLMs, fast thinking by triggering direct answers often outperforms slow thinking after deliberative reasoning.
- ▪Our empirical analyses show that fast thinking improves recall with broader and more confident predictions, whereas slow thinking favors precision through conservative filtering of incorrect categories.
- ▪Building on these insights, we propose MER-R1, a reinforcement learning framework that turns slow-fast complementarity into explicit optimization.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2606.27652 (cs) [Submitted on 26 Jun 2026] Title:MER-R1: Multimodal Emotion Reasoning via Slow-Fast Thinking Synergy Authors:Zhiyuan Han, Beier Zhu, Wenwen Tong, Chengwei Qin, Xinyi Wang, Jiayu Zhang, Jiangnan Chen, Hewei Guo, Dongchuan Ran, Lewei Lu, Xun Yang View a PDF of the paper titled MER-R1: Multimodal Emotion Reasoning via Slow-Fast Thinking Synergy, by Zhiyuan Han and 10 other authors View PDF HTML (experimental) Abstract:We find that explicit reasoning does not necessarily translate into better multimodal emotion recognition (MER) accuracy, even though it makes predictions more interpretable.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv.org.
Discussion
0 commentsMore from arXiv.org
