Abstract
Time Series Forecasting (TSF) is a crucial task in various domains, yetexisting TSF models rely heavily on high-quality data and insufficientlyexploit all available data. This paper explores a novel self-supervisedapproach to re-label time series datasets by inherently constructing candidatedatasets. During the optimization of a simple reconstruction network,intermediates are used as pseudo labels in a self-supervised paradigm,improving generalization for any predictor. We introduce the Self-Correctionwith Adaptive Mask (SCAM), which discards overfitted components and selectivelyreplaces them with pseudo labels generated from reconstructions. Additionally,we incorporate Spectral Norm Regularization (SNR) to further suppressoverfitting from a loss landscape perspective. Our experiments on elevenreal-world datasets demonstrate that SCAM consistently improves the performanceof various backbone models. This work offers a new perspective on constructingdatasets and enhancing the generalization of TSF models through self-supervisedlearning.