{"ID":2860191,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.04020","arxiv_id":"2510.04020","title":"Spatiotemporal Forecasting as Planning: A Model-Based Reinforcement Learning Approach with Generative World Models","abstract":"To address the dual challenges of inherent stochasticity and non-differentiable metrics in physical spatiotemporal forecasting, we propose Spatiotemporal Forecasting as Planning (SFP), a new paradigm grounded in Model-Based Reinforcement Learning. SFP constructs a novel Generative World Model to simulate diverse, high-fidelity future states, enabling an \"imagination-based\" environmental simulation. Within this framework, a base forecasting model acts as an agent, guided by a beam search-based planning algorithm that leverages non-differentiable domain metrics as reward signals to explore high-return future sequences. These identified high-reward candidates then serve as pseudo-labels to continuously optimize the agent's policy through iterative self-training, significantly reducing prediction error and demonstrating exceptional performance on critical domain metrics like capturing extreme events.","short_abstract":"To address the dual challenges of inherent stochasticity and non-differentiable metrics in physical spatiotemporal forecasting, we propose Spatiotemporal Forecasting as Planning (SFP), a new paradigm grounded in Model-Based Reinforcement Learning. SFP constructs a novel Generative World Model to simulate diverse, high-...","url_abs":"https://arxiv.org/abs/2510.04020","url_pdf":"https://arxiv.org/pdf/2510.04020v3","authors":"[\"Hao Wu\",\"Yuan Gao\",\"Xingjian Shi\",\"Shuaipeng Li\",\"Fan Xu\",\"Fan Zhang\",\"Zhihong Zhu\",\"Weiyan Wang\",\"Xiao Luo\",\"Kun Wang\",\"Xian Wu\",\"Xiaomeng Huang\"]","published":"2025-10-05T03:57:38Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.AI\"]","methods":"[\"Reinforcement Learning\"]","has_code":false}