Can I read ReckOn: A 28nm Sub-mm2 Task-Agnostic Spiking Recurrent Neural Network Processor Enabling On-Chip Learning over Second-Long Timescales on EtoBox?

ReckOn: A 28nm Sub-mm2 Task-Agnostic Spiking Recurrent Neural Network Processor Enabling On-Chip Learning over Second-Long Timescales by Frenkel, Charlotte; Indiveri, Giacomo is a scholarly article available to read on EtoBox.

What is ReckOn: A 28nm Sub-mm2 Task-Agnostic Spiking Recurrent Neural Network Processor Enabling On-Chip Learning over Second-Long Timescales about?

A robust real-world deployment of autonomous edge devices requires on-chip adaptation to user-, environment- and task-induced variability. Due to on-chip memory constraints, prior learning devices were limited to static stimuli with no temporal contents. We propose a 0.45-mm$^2$ spiking RNN processor enabling task-agnostic online learning over seconds, which we demonstrate for navigation, gesture recognition, and keyword spotting within a 0.8-% memory overhead and a <150-$\mu$W training power budget.

Author

Frenkel, Charlotte; Indiveri, Giacomo

Published

2022

Language

EN