# Algorithm and Hardware Design of Discrete-Time Spiking Neural Networks Based on Back Propagation with Binary Activations

arXiv: https://arxiv.org/abs/1709.06206 · HTML: https://mlsanity.com/paper/1709.06206/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 19 Sep 2017.

## Built on by (most cited first, 7 of 7)

- [Signal Propagation: A Framework for Learning and Inference In a Forward Pass](https://mlsanity.com/paper/2204.01723/) (2022)
- [Contrastive-Signal-Dependent Plasticity: Self-Supervised Learning in Spiking Neural Circuits](https://mlsanity.com/paper/2303.18187/) (2023)
- [Sparse Compressed Spiking Neural Network Accelerator for Object Detection](https://mlsanity.com/paper/2205.00778/) (2022)
- [Ultra-low-power Image Classification on Neuromorphic Hardware](https://mlsanity.com/paper/2309.16795/) (2023)
- [Neuromorphic Online Learning for Spatiotemporal Patterns with a Forward-only Timeline](https://mlsanity.com/paper/2307.11314/) (2023)
- [Advancing Neuromorphic Computing: Mixed-Signal Design Techniques Leveraging Brain Code Units and Fundamental Code Units](https://mlsanity.com/paper/2403.11563/) (2024)
- [A Differentiable Approach to Multi-scale Brain Modeling](https://mlsanity.com/paper/2406.19708/) (2024)

## Abstract (arXiv)

We present a new back propagation based training algorithm for discrete-time spiking neural networks (SNN). Inspired by recent deep learning algorithms on binarized neural networks, binary activation with a straight-through gradient estimator is used to model the leaky integrate-fire spiking neuron, overcoming the difficulty in training SNNs using back propagation. Two SNN training algorithms are proposed: (1) SNN with discontinuous integration, which is suitable for rate-coded input spikes, and (2) SNN with continuous integration, which is more general and can handle input spikes with temporal information. Neuromorphic hardware designed in 40nm CMOS exploits the spike sparsity and demonstrates high classification accuracy (>98% on MNIST) and low energy (48.4-773 nJ/image).