通过双流强化学习，让医疗多模态模型只关注关键图像区域，提速提效。
Token-Sparse Medical Multimodal Reasoning via Dual-Stream Reinforcement Learning
- 用双分支强化学习，一边定位关键图像区域，一边做稀疏推理。
- 在7个医疗数据集上将图像标记数减少77%，性能提升超100%。
- 适合追求高效医疗视觉推理的科研与临床应用者。
融合强化学习的视觉语言模型在多模态推理中取得显著进展，但在医疗图像场景中仍表现不足，因医学图像通常仅含极少的视觉证据支持临床决策。我们发现，剔除非关键区域的视觉标记可显著提升医疗推理效果。然而，尚无统一的强化学习框架实现主动视觉标记剪枝（VTP）与多模态推理协同优化。为此，我们提出双流强化学习框架ViToS，通过一个共享策略模型的双任务分支，分别聚焦于视觉定位与剪枝后的稀疏推理。为解决策略耦合问题，引入跨反馈序列优化机制，避免梯度冲突并促进收敛。在七个医疗基准上评估，该方法将视觉标记长度压缩至原长的77%，在Lingshu-7B上实现108.27%相对性能提升，在HuatuoGPT-Vision-7B上达104.16%相对提升。整体表现更优且推理速度加快，建立了高效的医疗多模态推理新范式。
原文摘要 · Abstract (English)
Vision-language models (VLMs) combining reinforcement learning (RL) ignite remarkable progress in multimodal reasoning, yet still struggle with medical images, which typically exhibit extremely sparse visual evidence to inform clinical decision-making. We recognize that pruning visual tokens outside the grounding region greatly enhances medical reasoning. However, a united RL framework for active visual token pruning (VTP) and medical multimodal reasoning remains unestablished. Here, we propose a dual-stream RL framework, ViToS, to fulfill token pruning and question answering. ViToS trains one policy model with two task branches, where one focuses on grounding while the other conducts token-sparse reasoning after VTP. Furthermore, we solve the coupled policy learning problem by introducing the cross-feedback sequential optimization, avoiding gradient conflict and facilitating convergence of the shared policy model. Evaluated on seven medical benchmarks, our method reduces visual tokens to 77% of the original sequence length while achieving a 108.27% relative performance on Lingshu-7B and 104.16% relative performance on HuatuoGPT-Vision-7B. Overall, ViToS delivers superior performance and inference speedup, establishing an efficient paradigm for medical multimodal reasoning.
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。