Self-Supervised Learning of Time Series Representation via Diffusion Process and Imputation-Interpolation-Forecasting Mask

arXiv:2405.05959v1 Announce Type: new
Abstract: Time Series Representation Learning (TSRL) focuses on generating informative representations for various Time Series (TS) modeling tasks. Traditional Self-Supervised Learning (SSL) methods in TSRL fall into four main categories: reconstructive, adversarial, contrastive, and predictive, each with a common challenge of sensitivity to noise and intricate data nuances. Recently, diffusion-based methods have shown advanced generative capabilities. However, they primarily target specific application scenarios like imputation and forecasting, leaving a gap in leveraging diffusion models for generic …

abstract adversarial arxiv challenge cs.ai cs.lg diffusion forecasting imputation interpolation modeling noise predictive process representation representation learning self-supervised learning sensitivity series ssl supervised learning tasks time series type via