Spike Processing on an Embedded Multi-task Computer: Image Reconstruction

Abstract

There is an emerging philosophy, called Neuro-informatics, contained in the Artificial Intelligence field, that aims to emulate how living beings do tasks such as taking a decision based on the interpretation of an image by emulating spiking neurons into VLSI designs and, therefore, trying to re-create the human brain at its highest level. address-event-representation (AER) is a communication protocol that has embedded part of the processing. It is intended to transfer spikes between bioinspired chips. An AER based system may consist of a hierarchical structure with several chips that transmit spikes among them in real-time, while performing some processing. There are several AER tools to help to develop and test AER based systems. These tools require the use of a computer to allow the higher level processing of the event information, reaching very high bandwidth at the AER communication level. We propose the use of an embedded platform based on a multi-task operating system to allow both, the AER communication and processing without the requirement of either a laptop or a computer. In this paper, we present and study the performance of a new philosophy of a frame-grabber AER tool based on a multi-task environment. This embedded platform is based on the Intel XScale processor which is governed by an embedded GNU/Linux system. We have connected and programmed it for processing Address-Event information from a spiking generato