Overview
Watch a technical conference talk from the tinyML Summit 2023 exploring how spiking neural networks (SNNs) can revolutionize sensor-edge computing. Learn about the brain-inspired computing paradigm that uses precisely-timed voltage pulses to process sensory information, and discover how Innatera's Spiking Neural Processor (SNP) implements these principles in an analog-mixed signal processing platform. Explore the PyTorch-compatible Talamo SDK for developing SNN applications, and see a practical demonstration of acoustic scene classification achieving state-of-the-art accuracy with ultra-low power consumption of 100μW and just 50ms latency. Understand how this neuromorphic approach could define the next generation of tiny machine learning at the sensor edge through the detailed presentation by Petrut Bogdan, Neuromorphic Architect at Innatera.
Syllabus
Intro
Outline
Bringing intelligence to the sensor edge
Processing sensor data with Spiking Neural Networks
Spiking Neural Processor (SNP)
Innatera's powerful Software Development Kit - Talamo
Building a spiking neural network in Talamo
Audio scene classification (DCASE 2020)
Summary
Taught by
EDGE AI FOUNDATION