Mamba4Cast: Efficient Zero-Shot Time Series Forecasting with State Space Models

TL;DR

Mamba4Cast is a novel zero-shot time series forecasting model based on the Mamba architecture, capable of generalizing across diverse datasets without fine-tuning, offering fast inference and competitive accuracy.

Contribution

The paper introduces Mamba4Cast, a zero-shot foundation model trained on synthetic data that outperforms traditional models in speed and scalability for time series forecasting.

Findings

Achieves strong zero-shot performance on real-world datasets

Offers significantly lower inference times than transformer-based models

Scales better with longer prediction horizons

Abstract

This paper introduces Mamba4Cast, a zero-shot foundation model for time series forecasting. Based on the Mamba architecture and inspired by Prior-data Fitted Networks (PFNs), Mamba4Cast generalizes robustly across diverse time series tasks without the need for dataset specific fine-tuning. Mamba4Cast's key innovation lies in its ability to achieve strong zero-shot performance on real-world datasets while having much lower inference times than time series foundation models based on the transformer architecture. Trained solely on synthetic data, the model generates forecasts for entire horizons in a single pass, outpacing traditional auto-regressive approaches. Our experiments show that Mamba4Cast performs competitively against other state-of-the-art foundation models in various data sets while scaling significantly better with the prediction length. The source code can be accessed at https://github.com/automl/Mamba4Cast.