面向城市时空预测的分布感知在线持续学习
Distribution-aware Online Continual Learning for Urban Spatio-Temporal Forecasting
AI总结：
针对城市时空预测中离线模型忽视数据分布随时间偏移导致性能下降的问题，提出DOST在线持续学习框架，通过自适应时空网络、醒睡学习策略及流记忆更新机制，在四个数据集上实现优于SOTA的预测性能，平均耗时0.1秒内且误差降低12.89%。
摘要：
城市时空（ST）预测对智能调度、行程规划等各类城市应用至关重要。以往研究聚焦于离线场景下城市区域间时空相关性的建模，却常忽视城市时空数据的非平稳特性，尤其是随时间发生的分布偏移，这一疏漏会导致其在实际场景中的性能下降。本文首先分析了城市时空数据中的分布偏移，随后提出DOST——一种针对时空数据特征定制的新型在线持续学习框架。DOST采用配备变量独立适配器的自适应时空网络，动态应对每个城市区域的独特分布偏移；为适应偏移的渐进特性，还开发了醒睡学习策略，在在线阶段间歇微调适配器以降低计算开销，该策略集成了针对城市时空序列数据的流记忆更新机制，能在防止灾难性遗忘的同时，有效让网络适应新模式。实验结果证实，DOST在四个真实世界数据集上优于现有最优模型，在线预测平均耗时0.1秒内，预测误差较基准模型降低12.89%。
英文摘要：
Urban spatio-temporal (ST) forecasting is crucial for various urban applications such as intelligent scheduling and trip planning. Previous studies focus on modeling ST correlations among urban locations in offline settings, which often neglect the non-stationary nature of urban ST data, particularly, distribution shifts over time. This oversight can lead to degraded performance in real-world scenarios. In this paper, we first analyze the distribution shifts in urban ST data, and then introduce DOST, a novel online continual learning framework tailored for ST data characteristics. DOST employs an adaptive ST network equipped with a variable-independent adapter to address the unique distribution shifts at each urban location dynamically. Further, to accommodate the gradual nature of these shifts, we also develop an awake-hibernate learning strategy that intermittently fine-tunes the adapter during the online phase to reduce computational overhead. This strategy integrates a streaming memory update mechanism designed for urban ST sequential data, enabling effective network adaptation to new patterns while preventing catastrophic forgetting. Experimental results confirm DOST's superiority over state-of-the-art models on four real-world datasets, providing online forecasts within an average of 0.1 seconds and achieving a 12.89% reduction in forecast errors compared to baseline models.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
