Find the workshop material on GitHub: https://github.com/synsense/snn-workshop-amld-2022

In this hands-on tutorial-style workshop, you’ll learn about spiking neural networks and how to use them to build real-time low-power machine learning applications. SynSense is a world-leading neuromorphic company which provides ultra-low-power ASICs for machine learning and signal processing. In this workshop our Algorithms team will talk about sensory processing inspired by the human retina and biological networks, machine learning algorithms that we use to train spiking neural networks, and our custom hardware that enables ultra-low-power applications. 

The full-day workshop overview:

• Present an introduction to spiking neural networks (SNNs) which perform highly energy-efficient machine learning and signal processing tasks.
• Present an overview of open-source tool chains for training and deploying SNN applications, based on PyTorch and JAX.
• Introduce neuromorphic sensing and processing HW, for real-time vision processing and auditory processing tasks.
• Provide a hands-on tutorial for every participant to execute on their own laptops. The task will be to fine-tune a pre-trained vision model for a gesture recognition task. We will talk about some of the intricacies when training SNNs and guide users through the process.
• Provide hands-on access to several development kits for vision and auditory processing, for attendees to test deployment of their models.
• Trained networks that do well on the test set can be deployed to our neuromorphic hardware kit that we provide on site. Such hardware kits comprise neuromorphic sensors as well as processing chips, which enable us to run live demos. 

At the end of the workshop you’ll be familiar with biologically-inspired spiking neural networks, a type of neural network that is getting traction in the domain of IoT and edge devices due to their low power consumption. You’ll have trained and deployed networks on SNN inference chips provided by SynSense, ready to be tested in live demos. Since all our tooling is open-source, attendees will be able to train their own networks using the Jupyter notebooks provided and the knowledge gained from the workshop.

Attendees should have trained a neural network before and be familiar with Python to follow along easily.

Find the workshop material on GitHub: https://github.com/synsense/snn-workshop-amld-2022