How Different from the Past? Spatio-Temporal Time Series Forecasting with Self-Supervised Deviation Learning

2025-10-06

AI Summary

To address insufficient dynamic bias modeling in spatiotemporal time series forecasting, this paper proposes a self-supervised bias learning framework. It discretizes the latent space via learnable prototypes to explicitly model dynamic deviations between current inputs and historical patterns. Two auxiliary objectives are introduced: a contrastive loss to enhance temporal consistency and a bias regularization loss to enforce discriminability of bias representations—both jointly optimized with the end-to-end forecasting task. This work is the first to incorporate self-supervised bias learning into spatiotemporal forecasting, enabling quantification and exploitation of bias signals without additional annotations. Extensive experiments on six benchmark datasets demonstrate consistent superiority over state-of-the-art methods across multiple metrics. Visualization analysis further confirms the framework’s strong adaptive responsiveness to biases of varying magnitudes.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Bias, Fairness & Privacy
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Spatio-temporal forecasting is essential for real-world applications such as traffic management and urban computing. Although recent methods have shown improved accuracy, they often fail to account for dynamic deviations between current inputs and historical patterns. These deviations contain critical signals that can significantly affect model performance. To fill this gap, we propose ST-SSDL, a Spatio-Temporal time series forecasting framework that incorporates a Self-Supervised Deviation Learning scheme to capture and utilize such deviations. ST-SSDL anchors each input to its historical average and discretizes the latent space using learnable prototypes that represent typical spatio-temporal patterns. Two auxiliary objectives are proposed to refine this structure: a contrastive loss that enhances inter-prototype discriminability and a deviation loss that regularizes the distance consistency between input representations and corresponding prototypes to quantify deviation. Optimized jointly with the forecasting objective, these components guide the model to organize its hidden space and improve generalization across diverse input conditions. Experiments on six benchmark datasets show that ST-SSDL consistently outperforms state-of-the-art baselines across multiple metrics. Visualizations further demonstrate its ability to adaptively respond to varying levels of deviation in complex spatio-temporal scenarios. Our code and datasets are available at https://github.com/Jimmy-7664/ST-SSDL.

Problem

• Captures dynamic deviations between current inputs and historical patterns
• Improves spatio-temporal forecasting accuracy using self-supervised learning
• Organizes latent space to handle diverse spatio-temporal input conditions

Innovation

• Self-supervised deviation learning captures input deviations
• Discretized latent space uses learnable spatio-temporal prototypes
• Contrastive and deviation losses refine hidden space structure