SpikeSTAG: Spatial-Temporal Forecasting via GNN-SNN Collaboration

2025-08-04

AI Summary

Spiking Neural Networks (SNNs) exhibit limited spatial modeling capability in multivariate time series forecasting. Method: This paper proposes the first spatio-temporal co-design architecture integrating Graph Neural Networks (GNNs) with SNNs. It innovatively unifies graph structure learning and spike-based temporal processing, introducing a multi-scale spike aggregation mechanism and a Dual-Path Spike Fusion (DSF) module. The DSF module incorporates learnable temporal embeddings, OBS-based sparsification, spike-adapted GraphSAGE layers, and an LSTM-spiking self-attention fusion mechanism—all operating within a fully spike-based framework to eliminate floating-point operations. Crucially, the model learns spatial dependencies adaptively without requiring predefined topologies. Results: Extensive experiments demonstrate that our method significantly outperforms the SNN baseline iSpikformer across multiple benchmark datasets and surpasses state-of-the-art conventional time series models on long-horizon forecasting tasks, validating its effectiveness and generalizability in joint spatio-temporal modeling.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

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

Abstract

Spiking neural networks (SNNs), inspired by the spiking behavior of biological neurons, offer a distinctive approach for capturing the complexities of temporal data. However, their potential for spatial modeling in multivariate time-series forecasting remains largely unexplored. To bridge this gap, we introduce a brand new SNN architecture, which is among the first to seamlessly integrate graph structural learning with spike-based temporal processing for multivariate time-series forecasting. Specifically, we first embed time features and an adaptive matrix, eliminating the need for predefined graph structures. We then further learn sequence features through the Observation (OBS) Block. Building upon this, our Multi-Scale Spike Aggregation (MSSA) hierarchically aggregates neighborhood information through spiking SAGE layers, enabling multi-hop feature extraction while eliminating the need for floating-point operations. Finally, we propose a Dual-Path Spike Fusion (DSF) Block to integrate spatial graph features and temporal dynamics via a spike-gated mechanism, combining LSTM-processed sequences with spiking self-attention outputs, effectively improve the model accuracy of long sequence datasets. Experiments show that our model surpasses the state-of-the-art SNN-based iSpikformer on all datasets and outperforms traditional temporal models at long horizons, thereby establishing a new paradigm for efficient spatial-temporal modeling.

Problem

• Bridging SNN potential for spatial modeling in time-series forecasting
• Integrating graph structural learning with spike-based temporal processing
• Improving model accuracy for long sequence datasets

Innovation

• Integrates graph learning with spike-based processing
• Uses Multi-Scale Spike Aggregation for feature extraction
• Combines spatial and temporal features via spike-gated fusion