Online Few-shot Gesture Learning on a Neuromorphic Processor

TL;DR

This paper introduces SOEL, a neuromorphic learning system that enables rapid online few-shot learning of new gestures using deep spiking neural networks on specialized hardware, combining transfer learning and neuroscience principles.

Contribution

The paper presents SOEL, a novel online learning system for neuromorphic processors that achieves fast adaptation to new classes with minimal updates, leveraging error-triggered learning.

Findings

SOEL enables rapid online adaptation to new gesture classes.

Deep SNNs on neuromorphic hardware can learn from streamed data.

Error-triggered updates improve learning efficiency.

Abstract

We present the Surrogate-gradient Online Error-triggered Learning (SOEL) system for online few-shot learning on neuromorphic processors. The SOEL learning system uses a combination of transfer learning and principles of computational neuroscience and deep learning. We show that partially trained deep Spiking Neural Networks (SNNs) implemented on neuromorphic hardware can rapidly adapt online to new classes of data within a domain. SOEL updates trigger when an error occurs, enabling faster learning with fewer updates. Using gesture recognition as a case study, we show SOEL can be used for online few-shot learning of new classes of pre-recorded gesture data and rapid online learning of new gestures from data streamed live from a Dynamic Active-pixel Vision Sensor to an Intel Loihi neuromorphic research processor.