|
Implementations of spiking neural networks (SNNs) on neuromorphic hardware promise orders of magnitude less power consumption than their non-spiking counterparts. The standard neuron model for spike-based computation on such systems has long been the leaky integrate-and-fire (LIF) neuron. Recent results have shown that the inclusion of biologically more realistic neuron dynamics through adaptive currents can boost the performance of SNNs on spatio-temporal processing tasks. The root of the superiority of these so-called adaptive LIF neurons however is not well understood.
In this talk, I will discuss the dynamical, computational, and learning properties of adaptive LIF neurons and networks thereof. I will also show that challenges related to stability and parameterization for this class of models can be effectively addressed, allowing to improve over state-of-the-art performances on common event-based benchmark datasets.
|
|