EmoGavel: Grounding-Aware Reinforcement Learning with Verifiable Rewards for Multimodal Emotion Preference Learning

Key points

• Multimodal emotion preference learning uses pairwise comparisons to identify which open-ended emotion descriptions align more closely with human judgments.1
• We first introduce grounded supervised fine-tuning (Grounded SFT), which exploits controlled-error construction information to jointly learn preference prediction, error localization, repair, and missing-evidence recovery.1
• We then optimize these verifiable outcomes with grounding-aware group relative policy optimization (GA-GRPO), combining preference-gated grounding rewards with dynamic zero-variance rescue to prioritize preference correctness and recover learning signals from all-wrong groups.1
• Experiments on EmoPrefer and controlled-error evaluation sets show that EmoGavel improves both natural preference prediction and controlled-error discrimination while maintaining strong error-grounding performance.1

Sources (1)

• [1]EmoGavel: Grounding-Aware Reinforcement Learning with Verifiable Rewards for Multimodal Emotion Preference Learning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 10:28 AM

  To address this limitation, we propose EmoGavel, a Grounding-Aware reinforcement learning framework with Verifiable rewards for multimodal Emotion preference Learning.

  Multimodal emotion preference learning uses pairwise comparisons to identify which open-ended emotion descriptions align more closely with human judgments.

Extractive summary: sentences quoted from the sources.