noisy-snn-learning

What problem does it solve?

Leverages noise as a computing resource to train and deploy noisy spiking neural networks (NSNN) with Noise-Driven Learning (NDL), enhancing robustness and probabilistic neural coding.

Core Features & Use Cases

• Noise-augmented SNN model: explicit stochastic dynamics for realistic neuromorphic behavior.
• NDL learning rule: built-in regularization and broader parameter exploration to improve generalization.
• Python reference implementation: modular components for neurons, layers, and training, with example workflows.
• Evaluation & visualization tools: utilities for robustness assessment and qualitative insights.

Quick Start

Run an end-to-end demonstration by calling the example_noisy_snn() function to instantiate, train, and visualize results.