Organizing a session at OSF2024

Abstract

The exponentially growing operational costs of modern AI systems are unsustainable in terms of energy and infrastructure requirements in the Netherlands. Neuromorphic computing with brain-inspired computing paradigms shows promise for extremely low-energy computing, thereby providing a greener and more sustainable AI solution. However, the microelectronic and hardware technology origins of neuromorphic computing create obstacles between pursuing open science and protecting intellectual property. This not only limits the reproducibility and fair comparison within the field, but also decelerates the research advancements with limited access to state-of-the-art neuromorphic hardware. Recent international collaborations in neuromorphic computing have started to transform the field to open hardware, software, and benchmarking. For example, the NeuroBench initiative targets to develop open benchmarks for neuromorphic algorithms and hardware that can fairly compare different neuromorphic solutions with reproducible results. Moreover, the Open Neuromorphic Community promoting open-sourced neuromorphic software and hardware designs and implementations has gradually gained more and more traction. In this session, we will introduce neuromorphic computing to the general public of the open science community and jointly discuss with the audiences the possible paths for advancing open neuromorphic initiatives. The session will start with an introduction to the history and advances of neuromorphic computing, starting from how to use digital and analog hardware to mimic biological neural computation, to cutting-edge compute-in-memory technologies, and brain-inspired learning algorithms. The introduction will provide the background to the follow-up panel discussion on the open neuromorphic initiatives. The panel discussion will invite top researchers from academia and industry to jointly discuss and interact with the audience about the future of open neuromorphic initiatives and how to create a more sustainable future in the wave of AI advancements.

Guangzhi Tang

Assistant Professor

Edge AI, Robotics, Neuromorphic Computing, and AI-enabled Automation