用段落级强化学习让AI看图说话更靠谱，提升深伪检测准确率。
PRPO: Paragraph-level Policy Optimization for Vision-Language Deepfake Detection
- 基于段落级相对策略优化，让大模型推理贴合视觉证据
- 检测准确率大幅提升，推理评分达4.55/5.0
- 适合关注可解释性深伪检测的研究者和安全应用
合成媒体的迅猛发展使深伪检测成为保障网络可信度的关键挑战。现有进展受限于高质量数据集稀缺。尽管多模态大语言模型具备强推理能力，但其在深伪检测中表现不佳，常生成与视觉证据不符或虚构的解释。为此，我们构建了一个带推理标注的深伪检测数据集，并提出段落级相对策略优化（PRPO）算法，通过强化学习在段落层面实现大模型推理与图像内容对齐。实验表明，PRPO显著提升检测准确率，在推理评分上达到4.55/5.0。消融实验进一步证明，PRPO在测试阶段显著优于GRPO。结果表明，将多模态推理扎根于视觉证据，是实现更可靠、可解释深伪检测的关键。
原文摘要 · Abstract (English)
The rapid rise of synthetic media has made deepfake detection a critical challenge for online safety and trust. Progress remains constrained by the scarcity of large, high-quality datasets. Although multimodal large language models (LLMs) exhibit strong reasoning capabilities, their performance on deepfake detection is poor, often producing explanations that are misaligned with visual evidence or hallucinatory. To address this limitation, we introduce a reasoning-annotated dataset for deepfake detection and propose Paragraph-level Relative Policy Optimization (PRPO), a reinforcement learning algorithm that aligns LLM reasoning with image content at the paragraph level. Experiments show that PRPO improves detection accuracy by a wide margin and achieves the highest reasoning score of 4.55/5.0. Ablation studies further demonstrate that PRPO significantly outperforms GRPO under test-time conditions. These results underscore the importance of grounding multimodal reasoning in visual evidence to enable more reliable and interpretable deepfake detection.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。