Resources

DCLL Library

Our project is based on the library for Deep Continuous Learning (DECOLLE) by Jacques Kaiser, Hesham Mostafa and Emre Neftei. DECOLLE is a novel learning technique for spiking neural networks which works often better on deep Spiking Neural Networks than previously proposed techniques.

Paper: Synaptic Plasticity Dynamics for Deep Continuous Local Learning

Code: Github

Brevitas

Brevitas is a Python library to assist with training quantized neural neworks developed by Xilinx.

It was developed for researchers to train for reduced precision inference.

Spiking Neural Networks

A description of the neuron model used in this project (SRM0) can be found in the book Spiking Neuron Models by Wulfram Gerstner and Werner M. Kistler.

The chapter is available online Spike Response Model (SRM).