SurF: A Generative Model for Multivariate Irregular Time Series Forecasting

TL;DR

SurF is a novel generative model for multivariate irregular time series that leverages the Time Rescaling Theorem, enabling effective training across diverse datasets and achieving state-of-the-art results on multiple benchmarks.

Contribution

It introduces SurF, a new model that uses TRT for flexible, scalable, and dataset-agnostic time series generation with Transformer-based pretraining.

Findings

SurF achieves the best time RMSE on Earthquake, Retweet, and Taobao datasets.

It outperforms classical and neural-autoregressive baselines on 5 out of 6 datasets.

SurF approaches the performance of specialized models on remaining datasets.

Abstract

Irregularly sampled multivariate event streams remain a stubbornly difficult modality for generative modeling: tokenization-based approaches break down when inter-event intervals vary by orders of magnitude, and neural temporal point processes are bottlenecked by window-level numerical quadrature. We (i) propose SurF, a generative model that uses the Time Rescaling Theorem (TRT) as a learnable bijection between event sequences and i.i.d.\ unit-rate exponential noise, enabling a single model to be trained across heterogeneous event-stream datasets; (ii) three efficient parameterizations of the cumulative intensity that scale to long sequences; and (iii) a Transformer-based encoder for multi-dataset pretraining. On six real-world benchmarks, SurF achieves the best reported time RMSE on Earthquake, Retweet, and Taobao, and is within trial-level noise of the strongest specialist on the remaining three. Under a strict leave-one-out protocol, the held-out checkpoint beats every classical and neural-autoregressive baseline on 5/6 datasets and beats every baseline on Amazon and Earthquake, an initial step toward foundation models over asynchronous event streams.