GeoCoT：面向可靠遥感推理的流形视角方法
GeoCoT: Towards Reliable Remote Sensing Reasoning with Manifold Perspective
提出基于局部流形映射的稀疏专家路由机制，可复用于视觉语言模型的多任务适配，降低高维视觉token冗余；同时引入RSV-GRPO强化结构化推理训练，为遥感视觉语言模型提供可借鉴的优化策略。
原文依据 GeoCoT introduces Mani-MoE, a sparse expert architecture grounded in local manifold mapping.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多模态大语言模型（MLLMs）通过多任务推理和跨模态泛化在遥感（RS）领域展现出强大潜力。然而，现有的遥感多模态大语言模型主要依赖单一共享专家处理所有任务，难以产生可靠结果。同时，遥感图像固有的冗余性和同质性给训练和推理都带来了巨大困难。这些挑战与遥感领域对任务精度和可信推理的需求直接冲突。为解决这些局限，我们提出GeoCoT，一个流形驱动的混合专家（MoE）系统，具备思维链（CoT）推理能力。GeoCoT引入了Mani-MoE，一种基于局部流形映射的稀疏专家架构。它将高维token自适应地投影到低秩子空间以消除冗余并揭示内在结构，然后通过稀疏专家路径进行路由，其中门控决策由输入的流形结构引导。为优化该架构，我们采用思维链驱动的多阶段训练策略。该策略利用冷启动阶段进行领域适应，随后采用我们的遥感视觉组相对策略优化（RSV-GRPO）从全局到目标系统地增强结构化推理。此外，我们创新性地构建了RS-CoT-20k数据集用于任务特定监督。在多任务数据集上的大量实验表明，GeoCoT优于先前方法，平均准确率比最优方法高出5.27%。我们的代码将公开。
英文原文摘要 1592 字符
Multimodal Large Language Models (MLLMs) have shown strong potential in remote sensing (RS) through multi-task reasoning and cross-modal generalization.However, existing RS-MLLMs mainly rely on a single shared expert for all tasks, making it hard to produce reliable results. Meanwhile, the intrinsic redundancy and homogeneity of RS images bring substantial difficulties for both training and inference. These challenges directly conflict with the demands of remote sensing, which values task precision and trustworthy reasoning.To address these limitations, we propose GeoCoT, a manifold-driven mixture-of-experts (MoE) system with Chain-of-Thought (CoT) reasoning. GeoCoT introduces Mani-MoE, a sparse expert architecture grounded in local manifold mapping. It projects high-dimensional tokens onto low-rank subspaces adaptively to eliminate redundancy and uncover intrinsic structure, and then routes them through a sparse expert pathway, where gating decisions are guided by the manifold structure of the input.To optimize this architecture, we adopt a CoT-driven multi-stage training strategy. It leverages a cold-start phase for domain adaptation, followed by our RS Vision Group Relative Policy Optimization (RSV-GRPO) to systematically strengthen structured reasoning from global to objectives. Furthermore, we innovatively build *RS-CoT-20k* dataset for task-specific supervision.Extensive experiments on multi-task datasets demonstrate that GeoCoT outperforms prior approaches, achieving 5.27 \% higher average accuracy than the state-of-the-art method. Our code will be available.