ODformer: Spatial–temporal transformers for long sequence Origin–Destination matrix forecasting against cross application scenario. (15th July 2023)

• Record Type: Journal Article
• Authors: Huang, Bosong; Ruan, Ke; Yu, Weihao; Xiao, Jing; Xie, Ruzhong; Huang, Jin
• Is Part Of: Expert systems with applications. Volume 222(2023)
• Publication Date: 2023-07-15
• Subjects: Origin–Destination matrix -- Graph convolutional network -- Sparse self-attention mechanism -- Long sequence time-series forecasting; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2023.119835
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: Origin–Destination (OD) matrices record directional flow data between pairs of OD regions. The intricate spatiotemporal dependency in the matrices makes the OD matrix forecasting (ODMF) problem not only intractable but also non-trivial. However, most of the related methods are designed for very short sequence time series forecasting in specific application scenarios, which cannot meet the requirements of the variation in scenarios and forecasting length of practical applications. To address these issues, we propose a Transformer-like model named ODformer, with two salient characteristics: (i) the novel OD Attention mechanism, which captures special spatial dependencies between OD pairs of the same origin (destination), greatly improves the ability of the model to predict cross application scenarios after combining with 2D-GCN that captures spatial dependencies between OD regions. (ii) a PeriodSparse Self-attention that effectively forecasts long sequence OD matrix series while adapting to the periodic differences in different scenarios. Generous experiments in three application backgrounds (i.e., transportation traffic, IP backbone network traffic, crowd flow) show that our method outperforms the state-of-the-art methods. Highlights: The first Transformer-like model for the Origin–Destination matrix forecasting. Crosses multiple application scenarios, covering three real-world applications. Origin–Destination attention mechanism to mine the specialized spatial dependency. Sparse Self-attention mechanism to capture long-range temporal correlation.