OmniST: Topology-Invariant Spatiotemporal Forecasting with Guaranteed-Coverage Uncertainty at Unseen Sensor Locations
作者: Praveg Vashishtha
分类: cs.LG, cs.AI, stat.ML
发布日期: 2026-11-03
备注: Code and experiments: https://github.com/Praveg432/OmniST . 9 tables, 4 figures. Baselines (E-STGCN, IGNNK) are faithful reimplementations, not the authors' original code
💡 一句话要点
提出OmniST以解决时空预测中传感器拓扑依赖问题
🎯 匹配领域: 支柱二:RL算法与架构 (RL & Architecture) 支柱八:物理动画 (Physics-based Animation)
关键词: 时空预测 拓扑不变性 图卷积网络 传感器网络 空气质量监测 归纳分裂保形校准 机器学习
📋 核心要点
- 现有的时空图卷积网络对传感器数量有固定依赖,无法处理缺失传感器的情况。
- OmniST通过坐标基础的标记化和k近邻注意力机制,实现了拓扑不变性,支持任意传感器位置的预测。
- 在多个真实数据集上,OmniST的预测性能显著优于邻近插值和学习克里金GNN,且具备有限样本覆盖保证。
📝 摘要(中文)
标准的时空图卷积网络需要固定大小的邻接矩阵,导致模型权重与特定传感器数量N绑定,无法在训练中缺失的传感器上产生输出。本文提出OmniST,这是一种拓扑不变的预测模型,其计算与节点数量无结构依赖,并结合归纳分裂保形校准,能够在任意(包括未见过的)传感器位置提供无分布、覆盖保证的预测区间。OmniST在真实的德里空气质量数据上表现出色,30天PM2.5的平均绝对误差为31.0微克/立方米,且在移除40%传感器时最多仅降幅1.1%。
🔬 方法详解
问题定义:本文旨在解决现有时空预测模型对传感器数量的固定依赖问题,导致在缺失传感器时无法进行有效预测。
核心思路:OmniST通过引入拓扑不变性,消除了对节点数量的结构依赖,利用坐标基础的标记化和k近邻注意力机制,使得模型能够在任意传感器位置进行预测。
技术框架:OmniST的整体架构包括坐标标记化模块、k近邻注意力机制、双向GRU进行节点的时间处理,以及归纳分裂保形校准模块,确保预测区间的覆盖性。
关键创新:OmniST的主要创新在于其拓扑不变性设计,使得模型在添加或移除传感器时仍能保持预测的有效性和覆盖保证,这与传统方法显著不同。
关键设计:模型采用了基于地理坐标的k近邻注意力机制,结合双向GRU进行时间序列处理,确保了每个节点的独立性和灵活性,同时在损失函数设计上考虑了覆盖保证的需求。
🖼️ 关键图片
📊 实验亮点
在德里空气质量数据集上,OmniST实现了30天PM2.5的平均绝对误差为31.0微克/立方米,并在移除40%传感器时仅降幅1.1%。其在未见传感器上的预测性能超越了邻近插值约12%和学习克里金GNN约21%,且具备90%以上的覆盖保证。
🎯 应用场景
OmniST在环境监测、交通流量预测和气象预报等领域具有广泛的应用潜力。其拓扑不变性和覆盖保证的特性使得在动态变化的传感器网络中,能够提供稳定可靠的预测,具有重要的实际价值和未来影响。
📄 摘要(原文)
Standard spatiotemporal graph convolutional networks need a fixed-size adjacency matrix, coupling model weights to a specific sensor count N; they cannot produce any output at a sensor absent from training. We present OmniST, a topology-invariant forecaster whose computation has zero structural dependency on the number of nodes, and which, composed with inductive split-conformal calibration, gives distribution-free, coverage-guaranteed prediction intervals at arbitrary (including never-seen) sensor locations, preserving that guarantee as sensors are added or removed. Topology invariance follows from coordinate-based tokenization, k-nearest-neighbor attention from geographic coordinates at each forward pass, and per-node temporal processing via a bidirectional GRU. On real Delhi air-quality data (37 stations, 2019-2023), OmniST attains a 30-day PM2.5 MAE of 31.0 ug/m3 and degrades at most 1.1 percent when 40 percent of stations are removed. Its inductive forecasts at unseen stations beat nearest-neighbor interpolation (about 12 percent) and a learned kriging GNN (IGNNK, about 21 percent), and, unlike either, carry a finite-sample coverage guarantee. The guarantee holds at 90 percent or above under 30-50 percent node dropout across four real datasets over three domains (Delhi and Beijing air quality, NOAA GHCN weather, Caltrans PeMSD7-M traffic), and still holds at 325 real sensors (full PeMS-BAY). We also fix a temporal-drift failure: static conformal under-covers over time (about 0.50 late), but a rolling adaptive-conformal estimator restores target coverage (0.89 overall, 0.96 late) at the cost of wider intervals.