SpikF-GO: Spiking Fourier Graph Operators for Multivariate Time Series Forecasting
Researchers have introduced SpikF-GO, a novel Spiking Neural Network (SNN) approach for multivariate time series forecasting. Unlike previous SNN methods that process variables independently, SpikF-GO models inter-variable dependencies by treating each observation as a graph node and incorporating spike-driven spectral processing. This method includes a learnable frequency gate and spiking neurons applied to Fourier components, aiming for energy efficiency and improved accuracy. Evaluations on eight benchmarks show SpikF-GO outperforming its artificial neural network counterpart, FourierGNN, and achieving a better average rank among SNN methods. AI
IMPACT Introduces a new method for multivariate time series forecasting using spiking neural networks, potentially offering energy-efficient alternatives.