E-prop on spinnaker 2: exploring online learning in spiking rnns on neuromorphic hardware
-
Updated
Feb 4, 2023 - Python
E-prop on spinnaker 2: exploring online learning in spiking rnns on neuromorphic hardware
Kinetic NMDA recurrent SNN with sparse e-prop
PG-LIF: a plateau-gated two-compartment spiking neuron for multi-timescale gradient propagation and local, hardware-compatible learning. Code and experiments for the Neurocomputing submission.
A from-scratch exploration of spiking neural networks, reservoir computing, and bio-plausible learning rules. The project is built with neuromorphic hardware constraints in mind.
To associate your repository with the e-prop topic, visit your repo's landing page and select "manage topics."