Research map / Large language models / Agents, code and math
ADEPT: RL-Aligned Agentic Decoding of Emotion via Evidence Probing Tools — From Consensus Learning to Ambiguity-Driven Emotion Reasoning
Esther Sun, Bo-Hao Su, Abinay Reddy Naini, Shinji Watanabe, Carlos Busso
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the ADEPT tool protocol to Qwen-3-Omni (without GRPO) (ablation) | Qwen-3-Omni · MSP-Podcast Test1 | Set Recall: 49.23 to 57.48 | Improved |
| add GRPO alignment to ADEPT (ablation) | ADEPT · MSP-Podcast Test1 | Primary Macro-F1: 35.71 -> 42.24 | Improved |
| add GRPO alignment to ADEPT (ablation) | ADEPT · MSP-Podcast Test1 | Jaccard: 44.50 -> 47.51 | Improved |
| add prior tools (ablation) | ADEPT · MSP-Podcast Test1 | Set Recall: 58.17 -> 61.92 | Improved |
| add semantic probing tools (ablation) | ADEPT · MSP-Podcast Test1 | Primary Macro-F1: 33.86 -> 42.24 | Improved |
| add acoustic probing tools (ablation) | ADEPT · MSP-Podcast Test1 | Jaccard: 41.28 -> 47.51 | Improved |
| add replay tool (ablation) | ADEPT · MSP-Podcast Test1 | Primary Macro-F1: 40.44 -> 42.24 | Improved |
| add Evidence Trust Gate (ablation) | ADEPT · MSP-Podcast Test1 | Jaccard: 44.96 -> 47.51 | Improved |
| ADEPT + GRPO versus fine-tuned WavLM | ADEPT · MSP-Podcast Test1 | Primary Macro-F1: 0.3640 -> 0.4224 | Improved |
| zero-shot ADEPT + GRPO versus supervised BLSP-Emo | ADEPT · IEMOCAP (zero-shot, mapped 7-way) | Set Recall: 52.30 -> 54.80 | Improved |
| zero-shot ADEPT + GRPO versus supervised BLSP-Emo | ADEPT · IEMOCAP (zero-shot, mapped 7-way) | Accuracy: 76.13 -> 54.83 | Hurt |
Closest papers by meaning
- EmotionThinker: Prosody-Aware Reinforcement Learning for Explainable Speech Emotion Reasoning (ICLR 2026 · similarity 0.88)
- EmoThinker: Advancing Visual-Acoustic Emotion Analysis via Structural Token Selection and Chain-of-Thought Reasoning (CVPR 2026 · similarity 0.87)
- MME-Emotion: A Holistic Evaluation Benchmark for Emotional Intelligence in Multimodal Large Language Models (ICLR 2026 · similarity 0.86)
- Emotion-Coherent Reasoning for Multimodal LLMs via Emotional Rationale Verifier (AAAI 2026 · similarity 0.86)
- AVERE: Improving Audiovisual Emotion Reasoning with Preference Optimization (ICLR 2026 · similarity 0.86)
- RLVER: Reinforcement Learning with Verifiable Emotion Rewards for Empathetic Agents (ICLR 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
