Long-term Forecasting with TiDE: Time-series Dense Encoder

TL;DR

TiDE is a simple, fast MLP-based encoder-decoder model for long-term time-series forecasting that outperforms or matches complex models while being significantly faster, leveraging theoretical guarantees and empirical success.

Contribution

We introduce TiDE, a novel MLP-based model for long-term forecasting that combines simplicity, speed, and theoretical optimality, outperforming Transformer-based approaches.

Findings

TiDE matches or exceeds prior methods on benchmarks.

TiDE is 5-10x faster than Transformer models.

The linear analogue of TiDE achieves near-optimal error rates.

Abstract

Recent work has shown that simple linear models can outperform several Transformer based approaches in long term time-series forecasting. Motivated by this, we propose a Multi-layer Perceptron (MLP) based encoder-decoder model, Time-series Dense Encoder (TiDE), for long-term time-series forecasting that enjoys the simplicity and speed of linear models while also being able to handle covariates and non-linear dependencies. Theoretically, we prove that the simplest linear analogue of our model can achieve near optimal error rate for linear dynamical systems (LDS) under some assumptions. Empirically, we show that our method can match or outperform prior approaches on popular long-term time-series forecasting benchmarks while being 5-10x faster than the best Transformer based model.