Research map / Large language models / Agents, code and math
FactGuard: Agentic Video Misinformation Detection via Reinforcement Learning
Zehao Li, Hongwei Yu, Hao Jiang, Qiang Sheng, Yilong Xu, Baolong Bi, Yang Li, Zhenlong Yuan, Yujun Cai, Zhaoqi Wang
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 reinforcement learning (decision-aware GRPO) after SFT (ablation) | FactGuard · FakeSV | accuracy: 62.0 -> 79.3 | Improved |
| add agentic SFT before RL (ablation) | FactGuard · FakeSV | accuracy: 73.1 -> 79.3 | Improved |
| add tool-usage reward Rtool (ablation) | FactGuard · FakeSV | accuracy: 77.7 -> 79.3 | Improved |
| add risk-aware reward Rrisk (ablation) | FactGuard · FakeSV | accuracy: 78.5 -> 79.3 | Improved |
| add decision-aware RL to regulate tool use (ablation) | FactGuard · average FactProbe calls per sample | tool calls: 4-5 -> 1-2 | Improved |
| add FactProbe and ClipScout tools without training (ablation) | Qwen2.5-VL-7B · FakeSV | accuracy: 55.6 -> 57.6 | No clear effect |
| add FactProbe to FactGuard (ablation) | FactGuard · FakeSV | accuracy: 76.1 -> 79.3 | Improved |
| add ClipScout to FactGuard (ablation) | FactGuard · FakeSV | accuracy: 78.2 -> 79.3 | Improved |
| increase GRPO rollouts from 6 to 8 (ablation) | FactGuard · FakeSV | accuracy: 77.3 -> 79.3 | Improved |
| increase GRPO rollouts from 8 to 12 (ablation) | FactGuard · FakeSV | accuracy: 79.3 -> 79.5 | No clear effect |
| change KL coefficient from 0.04 to 0.02 or 0.06 (ablation) | FactGuard · FakeSV | accuracy: 79.3 -> 78.2 (0.02), 78.8 (0.06) | Hurt |
| swap teacher model from Qwen2.5-VL-72B to InternVL2.5-78B-MPO (ablation) | FactGuard SFT data · FakeSV | accuracy: 79.3 -> 78.7 after SFT+RL (62.0 -> 60.8 SFT only) | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Fact-R1 — Fact-R1: Towards Explainable Video Misinformation Detection with Deep Reasoning (NeurIPS 2025) · better on 7 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Fact-R1: Towards Explainable Video Misinformation Detection with Deep Reasoning (NeurIPS 2025 · similarity 0.93)
- Incentivizing Versatile Video Reasoning in MLLMs via Data-Efficient Reinforcement Learning (CVPR 2026 · similarity 0.86)
- Beyond Detection: Exploring Evidence-based Multi-Agent Debate for Misinformation Intervention and Persuasion (AAAI 2026 · similarity 0.86)
- Debate with Images: Detecting Deceptive Behaviors in Multimodal Large Language Models (ICML 2026 · similarity 0.86)
- SNIFFER: Multimodal Large Language Model for Explainable Out-of-Context Misinformation Detection (CVPR 2024 · similarity 0.85)
- Goal-Aware Identification and Rectification of Misinformation in Multi-Agent Systems (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.
