Byte Pair Encoding for Efficient Time Series Forecasting
This work addresses computational overhead in time series forecasting for applications using foundation models, representing a novel method rather than an incremental improvement.
The paper tackled the inefficiency of existing time series tokenization methods by proposing a pattern-centric tokenization scheme based on byte pair encoding, which improved forecasting performance by 36% and efficiency by 1990% on average, with conditional decoding further reducing MSE by up to 44%.
Existing time series tokenization methods predominantly encode a constant number of samples into individual tokens. This inflexible approach can generate excessive tokens for even simple patterns like extended constant values, resulting in substantial computational overhead. Inspired by the success of byte pair encoding, we propose the first pattern-centric tokenization scheme for time series analysis. Based on a discrete vocabulary of frequent motifs, our method merges samples with underlying patterns into tokens, compressing time series adaptively. Exploiting our finite set of motifs and the continuous properties of time series, we further introduce conditional decoding as a lightweight yet powerful post-hoc optimization method, which requires no gradient computation and adds no computational overhead. On recent time series foundation models, our motif-based tokenization improves forecasting performance by 36% and boosts efficiency by 1990% on average. Conditional decoding further reduces MSE by up to 44%. In an extensive analysis, we demonstrate the adaptiveness of our tokenization to diverse temporal patterns, its generalization to unseen data, and its meaningful token representations capturing distinct time series properties, including statistical moments and trends.