用流匹配直接建模事件时间联合分布，预测更准且更快。
EventFlow: Forecasting Temporal Point Processes with Flow Matching
- 不依赖自回归，用流匹配直接学习事件时间分布
- 在标准基准上误差降低20%-53%，采样调用次数更少
- 适合需要长时序预测的工业级事件建模场景
连续时间事件序列在众多工业与科学领域中普遍存在，其事件发生时间不规则。当前建模范式将此类数据视为时序点过程（Temporal Point Process, TPP）的实现，并常采用神经网络进行自回归建模。尽管自回归模型能较好预测单个后续事件的时间，但在长时程预测中因误差累积和短视预测导致性能下降。本文提出EventFlow，一种非自回归生成模型，基于流匹配框架，直接学习事件时间的联合分布，跳过自回归过程。该模型实现简单，在标准TPP基准上相较最近基线模型预测误差降低20%-53%，同时采样阶段调用次数更少。
原文摘要 · Abstract (English)
Continuous-time event sequences, in which events occur at irregular intervals, are ubiquitous across a wide range of industrial and scientific domains. The contemporary modeling paradigm is to treat such data as realizations of a temporal point process, and in machine learning it is common to model temporal point processes in an autoregressive fashion using a neural network. While autoregressive models are successful in predicting the time of a single subsequent event, their performance can degrade when forecasting longer horizons due to cascading errors and myopic predictions. We propose EventFlow, a non-autoregressive generative model for temporal point processes. The model builds on the flow matching framework in order to directly learn joint distributions over event times, side-stepping the autoregressive process. EventFlow is simple to implement and achieves a 20%-53% lower forecast error than the nearest baseline on standard TPP benchmarks while simultaneously using fewer model calls at sampling time.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。