AutoSTF: Decoupled Neural Architecture Search for Cost-Effective Automated Spatio-Temporal Forecasting

2024-09-25

arXiv.org

AI Summary

Existing neural architecture search (NAS) methods for spatiotemporal forecasting in smart cities suffer from high computational overhead and limited support for fine-grained spatiotemporal operator exploration. Method: We propose a decoupled NAS framework featuring: (i) a novel spatiotemporal decoupling search paradigm that separates spatial and temporal modeling; (ii) representation compression and parameter sharing to reduce search cost; (iii) multi-block transfer modules for multi-granularity temporal modeling; and (iv) a layer-level fine-grained spatial search space. The method integrates spatiotemporal graph neural networks with differentiable NAS optimization. Contribution/Results: Our approach achieves state-of-the-art accuracy on eight benchmark datasets while accelerating the search process by up to 13.48×. It significantly reduces computational cost without compromising predictive performance, enabling efficient and scalable architecture discovery for urban spatiotemporal forecasting.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Learning to Search
• Machine Learning: Deep Neural Architectures and Foundation Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Spatio-temporal forecasting is a critical component of various smart city applications, such as transportation optimization, energy management, and socio-economic analysis. Recently, several automated spatio-temporal forecasting methods have been proposed to automatically search the optimal neural network architecture for capturing complex spatio-temporal dependencies. However, the existing automated approaches suffer from expensive neural architecture search overhead, which hinders their practical use and the further exploration of diverse spatio-temporal operators in a finer granularity. In this paper, we propose AutoSTF, a decoupled automatic neural architecture search framework for cost-effective automated spatio-temporal forecasting. From the efficiency perspective, we first decouple the mixed search space into temporal space and spatial space and respectively devise representation compression and parameter-sharing schemes to mitigate the parameter explosion. The decoupled spatio-temporal search not only expedites the model optimization process but also leaves new room for more effective spatio-temporal dependency modeling. From the effectiveness perspective, we propose a multi-patch transfer module to jointly capture multi-granularity temporal dependencies and extend the spatial search space to enable finer-grained layer-wise spatial dependency search. Extensive experiments on eight datasets demonstrate the superiority of AutoSTF in terms of both accuracy and efficiency. Specifically, our proposed method achieves up to 13.48x speed-up compared to state-of-the-art automatic spatio-temporal forecasting methods while maintaining the best forecasting accuracy.

Problem

• Automated Model Optimization
• Computational Resource Consumption
• Spatio-Temporal Prediction

Innovation

• AutoSTF
• spatio-temporal prediction
• model optimization