Introducing spikeDE: Where Fractional Calculus Meets Spiking Neural Networks
Today marks a significant milestone for our research team. We are incredibly proud to announce the public release of spikeDE, an open-source PyTorch based library designed to bring Fractional-Order Dynamics to the world of Spiking Neural Networks (SNNs).
For years, SNNs have been celebrated for their biological plausibility and energy efficiency. However, traditional models like the Leaky Integrate-and-Fire (LIF) neuron rely on integer-order differential equations (\(\alpha=1\)), which assume Markovian dynamics—meaning the neuron's current state depends only on its immediate past. This simplification fails to capture the rich, complex temporal dependencies observed in real biological neurons, which exhibit long-term memory and power-law relaxation.
With spikeDE, we change the paradigm.