SpikF-GO: Spiking Fourier Graph Operators for Multivariate Time Series Forecasting

In three linesSpikF-GO combines Spiking Neural Networks (SNNs) with Fourier graph operators for multivariate time series forecasting. The method introduces Hard Concrete frequency gates and Complex LIF gates to process real and imaginary Fourier components, reducing energy consumption while outperforming FourierGNN across 8 benchmarks.

Summary generated by Claude — human-verified