Introducing the Dendrify framework for incorporating dendrites to spiking neural networks

Spiking neural networks (SNNs) are widely used for understanding brain functions and neuromorphic computing, but often ignore dendrites' computational capabilities. Dendrify, an open-source Python package based on Brian 2, generates reduced compartmental neuron models with biologically relevant dendritic and synaptic properties. This bridges the gap between efficient but simplistic SNNs and computationally expensive detailed models, allowing exploration of dendritic contributions to network-level functions and advancing neuromorphic systems.

Michalis Pagkalos, Spyridon Chavlis, Panayiota Poirazi