TimePFN: Effective Multivariate Time Series Forecasting with Synthetic Data

TL;DR

TimePFN introduces a transformer-based approach that leverages synthetic data priors for multivariate time series forecasting, achieving strong zero- and few-shot performance with minimal data.

Contribution

The paper proposes a novel training scheme and architecture, combining synthetic data generation with a transformer model for improved multivariate time series forecasting.

Findings

Outperforms state-of-the-art models in zero- and few-shot settings

Nearly matches full dataset training error with only 500 data points

Demonstrates strong univariate forecasting performance

Abstract

The diversity of time series applications and scarcity of domain-specific data highlight the need for time-series models with strong few-shot learning capabilities. In this work, we propose a novel training scheme and a transformer-based architecture, collectively referred to as TimePFN, for multivariate time-series (MTS) forecasting. TimePFN is based on the concept of Prior-data Fitted Networks (PFN), which aims to approximate Bayesian inference. Our approach consists of (1) generating synthetic MTS data through diverse Gaussian process kernels and the linear coregionalization method, and (2) a novel MTS architecture capable of utilizing both temporal and cross-channel dependencies across all input patches. We evaluate TimePFN on several benchmark datasets and demonstrate that it outperforms the existing state-of-the-art models for MTS forecasting in both zero-shot and few-shot settings. Notably, fine-tuning TimePFN with as few as 500 data points nearly matches full dataset training error, and even 50 data points yield competitive results. We also find that TimePFN exhibits strong univariate forecasting performance, attesting to its generalization ability. Overall, this work unlocks the power of synthetic data priors for MTS forecasting and facilitates strong zero- and few-shot forecasting performance.