If we compare the human brain to society’s current computation abilities, our minds still remain the most powerful central processing units in existence.
Inspired by the brains exceptional complexity and processing power, computer scientist and engineer Carver Mead developed the concept of Neuromorphic Computing
in the late 1980’s.
What is Neuromorphic Computing?
Neuromorphic Computing explores the potential for modelling biological systems of computation on both animal and human brains. It involves the use of very-large-scale integration (VLSI) systems constructed of electronic analog circuits to mimic neurobiological architectures present in the human nervous system. In the neuromorphic process chips encode and transmit data in a way that mimics the electrical spikes generated in the brain as it responds to sensory information.
Why Is It Trending?
Modern computation as we know it was first outlined by the mathematician John von Neumann and his colleagues in 1945. In a von Neumann machine, all data-crunching happens in the central processing unit (C.P.U.). The program instructions, followed by the data, flows from a computer’s memory to its C.P.U. in an orderly series of zeroes and ones, in similarity to a stack of punch cards shuffling through. A modern laptop is conceptually identical to the punch-card behemoths of the past, although engineers have traded paper for a purely electric stream of on-off signals. Until now Von Neumman architecture has allowed a development in continuous technological improvements as predicted by Moore’s Law
. However, due to fundamental scientific reasons, computation using the current architectures and materials available today will inevitably reach a limit within the next 10 years.
So What’s The Solution?
Enter parallel processing.
Certain multicore computers allow for parallel processing, however, their efficacy is limited. The process requires arduous choreography of information streams by software engineers in order to avoid catastrophic system errors. By contrast, in neural activity in the brain, data is able to run simultaneously via billions of parallel processors. In similarity to computers, communication occurs through a binary language of electrical spikes; however in difference, either through genetic patterning or via learned associations, each neuron is pre-programmed to share its computations with direct targets. The brain does not require a C.P.U as processing occurs organically.
The Brains in Silicon
Developments in Neuromorphic Computing
team at Stanford University have built Neurogrid
, an example of hardware designed using neuromorphic-engineering principles. The circuit board is composed of 16 custom-designed chips, which are referred to as NeuroCores. Each NeuroCore’s analog circuitry maximises energy efficiency by emulating neural elements for 65536 neurons. These emulated neurons are designed to maximise spiking throughput and are connected via digital circuitry.
Another research project incorporating neuromorphic engineering is the Human Brain Project
, a 10-year collaboration with the aim to simulate a complete human brain in a supercomputer using biological data. Henry Markram, the project’s co-director plans to create a new foundation with three primary goals:
To better understand how the pieces of the brain fit and work together
To understand how to objectively diagnose and treat brain disorders
Use the understanding of the human brain to develop neuromorphic computers.
A complete simulation of the human brain would require a supercomputer a thousand times more powerful than what we currently have on offer today. The European Commission has allocated $1.3 billion to the project. Other research with inferences for neuromorphic engineering involves the BRAIN Initiative
and the TrueNorth
chip from IBM.
Application in a Consumer Market
Need a fitness tracker that recognizes your every move and doesn’t require you to upload any information to a separate computer for analysis? Or maybe an auto-translator with an always-on Internet connection that can decipher any foreign speech within your general vicinity. Perhaps a self-driving car that can navigate effectively off grid or a version of Siri that gets it right more often than not.
Want To Know More?
All of the above are example of what neuromorphic chips could bring into our lives. In the meantime, want to learn more about high performance computing? Check out what we do here
and get in contact at firstname.lastname@example.org or +61 2 8572 4700 for more information.