This repository supports our research on the energy consumption of Spiking Neural Networks (SNNs), with a focus on:
- 📈 Spiking activity analysis – how energy varies with different levels of spiking across various network architectures
- 🔧 Spike budget-aware training – methods to reduce dynamic energy consumption by controlling spike activity during training
The goal is to better understand and optimize the energy efficiency of SNNs, especially for resource-constrained edge devices.
Updates:
- Spiking Energy and Timesteps analysis is been done using Nengo Loihi Emulator, it's theoretical estimation based on modeling loihi hardware. The complete visualisations are available in the Notion link