We introduce SiamTST, a novel representation learning framework for multivariate time series. SiamTST integrates a Siamese network with attention, channel-independent patching, and normalization techniques to achieve superior performance. Evaluated on a real-world industrial telecommunication dataset, SiamTST demonstrates significant improvements in forecasting accuracy over existing methods. Notably, a simple linear network also shows competitive performance, achieving the second-best results, just behind SiamTST. The code is available at https://github.com/simenkristoff/SiamTST.
@article{arxiv.2407.02258,
title = {SiamTST: A Novel Representation Learning Framework for Enhanced Multivariate Time Series Forecasting applied to Telco Networks},
author = {Simen Kristoffersen and Peter Skaar Nordby and Sara Malacarne and Massimiliano Ruocco and Pablo Ortiz},
journal= {arXiv preprint arXiv:2407.02258},
year = {2024}
}