Neuromorphic Computing Journey (Part 2)
Welcome again!
Thank you for the reading first part of this series. In this part, I will discuss the Human-Brain Project which is the flagship project of the European Union. Many quality pieces of research have been conducted under this project. Moreover, we will also look into the SpiNNaker project and its contribution to the field of Neuromorphic computing.
Human-Brain Project
This project aims to put in place a cutting-edge research infrastructure that will allow scientific and industrial researchers to advance our knowledge in the fields of neuroscience, computing and the brain-related medicine.
The focus of our journey is towards Robotics side of Human-Brain Project, so now will talk about HBP contribution in the field of neurorobotics. As discussed in previous part, the human brain is one of the most astonishing, incredibly complex and powerful creation of nature. Our brain makes body perform an action, then body perceive the results of this action and again the brain interprets the results to change the behaviour accordingly, so that next action can be more effective.
HBP introduced Neurorobotics Platform which provides a simulated environment for neuro-robots. Projects build on this platform show that these robots can construct their own effective and powerful learning rules, almost like living creatures. This platform is public, online, and available for all researchers who want to test their brain models or to build the brain-inspired robots of the future.
SpiNNaker: Multi-core System on Chip
The massive HBP project is divided into a number of different themes called pillars. The group in Manchester is the part of neuromorphic pillar whose interest is in developing and supporting novel computer hardware which can accelerate the simulation of large neural networks.
SpiNNaker is a novel computer architecture inspired by the human brain, which itself is composed of billion of simple computing elements, communicating using unreliable spikes. This project started in 2005, with funding from EPSRC, and with collaboration between several universities and industrial partners. SpiNNaker is a biological inspired massive parallel computing architecture designed to facilitate the modeling and simulation of large-scale spiking neural networks of up to a billion neurons and a trillion synapses in biological real-time. It is a general-purpose, programmable platform for neuroscientists, psychologists, and brain researchers to explore brain functions with software neural models. Don’t get confused with unfamiliar words here, I’ll explain them one by one. For now, just try to absorb them.
SpiNNaker is a good target for researchers in robotics, who need mobile, low power computation. A small SpiNNaker board makes it possible to simulate a network of tens of thousands of spiking neurons, process sensory input, and generate motor output, all in real-time and in a low power system.
How the question is, how to use this platform? Well, there are two options. 1) Get access to SpiNNaker via HBP portal through which you can submit jobs to the half-million core SpiNNaker machine.
2) Request to get a local SpiNNaker system for real-time interaction with a physical robotics system. So far, they are not commercially available but can only be requested to use for academic research. There are three options:
- 4-Node board (72 processors)
- 48-Node board (864 processors)
- 24-Board frame (20,736 processors)
Each chip on spiNNaker board contain 18 identical processing subsystems located in the periphery, and the Network-on-Chip and shared components in the centre. Inter-processor communication is based on an efficient multicast infrastructure inspired by neurobiology. It uses a packet-switched network to emulate the very high connectivity of biological systems. SpiNNaker chips have six bidirectional, inter-chip links that allow networks of various topologies. Inter-chip communication uses self-timed channels, which, although costly in wires, are significantly more power efficient than synchronous links of similar bandwidth.
For now, we are looking at these projects from far away, but in the future, we will have closer look at these platforms and their research contributions so far. Besides this, we will also look at how beginners in this field can use these platforms for their own projects.
That’s it for today. There is a possibility that you might have a lot of questions in your mind, but believe me, with this series you will have better knowledge of how things work in this Neuromorphic computing field. In the next part, I will discuss a little more on the programming side of the SpiNNaker board and then we will jump into Spiking-Neural Networks along with their similarity with the human brain.
Things will get exciting with time. Thanks for reading this blog.
Don’t forget to follow.
