Sequential Order-Robust Mamba for Time Series Forecasting

Seunghan Lee, Juri Hong, Kibok Lee, Taeyoung Park

Keywords: Time series, Mamba, State-space model, Self-supervised learning

TL;DR: We propose SOR-Mamba, a time series forecasting method that uses Mamba with regularization to address the sequential order bias in capturing channel dependencies.

Abstract:

Mamba has recently emerged as a promising alternative to Transformers, offering near-linear complexity in processing sequential data.
However, while channels in time series (TS) data have no specific order in general, recent studies have adopted Mamba to capture channel dependencies (CD) in TS, introducing sequential order bias. To address this issue, we propose SOR-Mamba, a TS forecasting method that 1) incorporates a regularization strategy to minimize the discrepancy between two embedding vectors generated from data with reversed channel orders, thereby enhancing robustness to channel order, and 2) eliminates the 1D-convolution originally designed to capture local information in sequential data. Furthermore, we introduce channel correlation modeling (CCM), a pretraining task designed to preserve correlations between channels from the data space to the latent space, thereby improving the ability to capture CD. Extensive experiments demonstrate the efficacy of the proposed method across standard and transfer learning scenarios.

Submission Number: 9