发表机构
Beihang University（北京航空航天大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
LLMODE是基于冻结LLM的框架，通过图感知ODE编码器等处理不规则时空数据，在多数据集上表现优异，零样本泛化能力强。
AI 中文摘要
大语言模型（LLM）在时空预测中展现出应用前景，但现有方法通常依赖规则采样的令牌序列，因时间异步性、表示空间错位及有限上下文窗口，难以处理不规则观测。我们提出LLMODE，一种基于冻结LLM骨干的令牌高效不规则时空预测框架。LLMODE首先使用图感知常微分方程（ODE）编码器将不规则图观测重构为连续时间潜轨迹；随后，固定预算感知器重采样器将该可变长度轨迹压缩为固定数量的动态记忆令牌；同时，紧凑统计描述符被编码并重采样为上下文记忆令牌；双源门控交叉注意力模块将两种记忆注入冻结LLM，实现对外部时空证据的可控利用。在3个真实城市数据集和2个物理动力学基准上的实验显示，其整体性能具竞争力，在稀疏或动态复杂不规则采样下优势更明显；对未见过的城市区域的额外评估进一步表明，其无需适配即可实现强零样本泛化。
英文摘要
Large language models (LLMs) have shown promise for spatio-temporal forecasting, but existing approaches often rely on regularly sampled token sequences and struggle with irregular observations because of temporal asynchrony, representation-space misalignment, and limited context windows. We propose LLMODE, a token-efficient framework for irregular spatio-temporal forecasting with a frozen LLM backbone. LLMODE first uses a graph-aware ODE encoder to reconstruct irregular graph observations as a continuous-time latent trajectory. A Fixed-Budget Perceiver Resampler then compresses this variable-length trajectory into a fixed number of dynamic memory tokens. In parallel, compact statistical descriptors are encoded and resampled into context memory tokens. A dual-source gated cross-attention module injects both memories into the frozen LLM, enabling controlled utilization of external spatio-temporal evidence. Experiments on three real-world urban datasets and two physical-dynamics benchmarks show competitive overall performance, with clearer advantages under sparse or dynamically complex irregular sampling. Additional evaluations on unseen urban regions further demonstrate strong zero-shot generalization without adaptation.
