Brain-Inspired Hardware for Artificial Intelligence: Accelerated Learning in a Physical-Model Spiking Neural Network

TL;DR

This paper presents a neuromorphic chip that accelerates neural dynamics by 1000 times, enabling efficient, noise-resilient reinforcement learning for tasks like playing Pong, demonstrating advances in brain-inspired AI hardware.

Contribution

The work introduces a physical-model neuromorphic system with accelerated dynamics and embedded plasticity, enabling on-chip reinforcement learning and demonstrating key brain-inspired computing features.

Findings

Achieved 1000-fold acceleration of neural dynamics.

Enabled on-chip reinforcement learning to play Pong.

Demonstrated high energy efficiency and noise resilience.

Abstract

Future developments in artificial intelligence will profit from the existence of novel, non-traditional substrates for brain-inspired computing. Neuromorphic computers aim to provide such a substrate that reproduces the brain's capabilities in terms of adaptive, low-power information processing. We present results from a prototype chip of the BrainScaleS-2 mixed-signal neuromorphic system that adopts a physical-model approach with a 1000-fold acceleration of spiking neural network dynamics relative to biological real time. Using the embedded plasticity processor, we both simulate the Pong arcade video game and implement a local plasticity rule that enables reinforcement learning, allowing the on-chip neural network to learn to play the game. The experiment demonstrates key aspects of the employed approach, such as accelerated and flexible learning, high energy efficiency and resilience to noise.