Unified training of universal time series forecasting transformers

• Main Authors: WOO, Gerald, LIU, Chenghao, KUMAR, Akshat, XIONG, Caiming, SAVARESE, Silvio, SAHOO, Doyen
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2024
• Subjects: Time series forecast; Deep learning; Time series transformer; Artificial Intelligence and Robotics
• Online Access: https://ink.library.smu.edu.sg/sis_research/9906; https://ink.library.smu.edu.sg/context/sis_research/article/10906/viewcontent/2402.02592v2.pdf
• Institution: Singapore Management University

Deep learning for time series forecasting has traditionally operated within a one-model-per-dataset framework, limiting its potential to leverage the game-changing impact of large pre-trained models. The concept of universal forecasting, emerging from pre-training on a vast collection of time series datasets, envisions a single Large Time Series Model capable of addressing diverse downstream forecasting tasks. However, constructing such a model poses unique challenges specific to time series data: i) cross-frequency learning, ii) accommodating an arbitrary number of variates for multivariate time series, and iii) addressing the varying distributional properties inherent in large-scale data. To address these challenges, we present novel enhancements to the conventional time series Transformer architecture, resulting in our proposed Masked Encoder-based Universal Time Series Forecasting Transformer (Moirai). Trained on our newly introduced Large-scale Open Time Series Archive (LOTSA) featuring over 27B observations across nine domains, Moirai achieves competitive or superior performance as a zero-shot forecaster when compared to full-shot models.