A Swinburne University of Technology demonstrated that the world’s fastest and most powerful optical neuromorphic artificial intelligence (AI) processor is capable of operating faster than 10 trillion operations per second and capable of processing large-scale data. The research published in the journal Nature represents a huge leap forward for neural networks and neuromorphic processing in general.
Artificial neural networks, a key form of artificial intelligence, can learn and perform complex operations with broad applications to computer vision, natural language processing, facial recognition, speech translation, strategy games, medical diagnosis, and many other areas. Inspired by the biological structure of the brain’s visual cortex system, artificial neural networks extract key features of raw data to predict properties and behavior with unprecedented accuracy and simplicity. Led by Professor David Moss of Swinburne, Dr Xingyuan (Mike) Xu (Swinburne, Monash University) and the distinguished Professor Arnan Mitchell from RMIT University, the team achieved an outstanding feat in optical neural networks: greatly accelerating processing speed and processing power.
The team demonstrated an optical neuromorphic processor that works more than 1000 times faster than any previous processor, with the system also processing record-sized images on an ultra-large scale enough to achieve full facial image recognition, something others optical processors have not been able to realize. “This breakthrough was achieved with ‘optical micro-combs’, as was our world-record Internet data rate reported in May 2020,” said Professor Moss, director of the Swinburne Optical Science Center and recently named one of the leading Australian research leaders in physics and mathematics in the field of optics and photonics from The Australian.
Although state-of-the-art electronic processors like Google TPU can run beyond 100 TeraOP / s, this is done with tens of thousands of parallel processors. In contrast, the optical system demonstrated by the team uses a single processor and was achieved using a new technique of simultaneous interlacing of data in time, wavelength and spatial dimensions through an integrated micro-comb source. Micro-combs are relatively new devices that behave like a rainbow made up of hundreds of high-quality infrared lasers on a single chip. They’re much faster, smaller, lighter and cheaper than any other optical source, according to the study.
Co-lead author of the study, Dr. Xu, a former Swinburne student and postdoctoral fellow in Monash University’s electrical and computer systems engineering department, said, “This processor can serve as a universal front-end wide high-bandwidth data for any optical or electronic neuromorphic hardware. Based on mass data machine learning for real-time ultra-high bandwidth data at your fingertips. “” We are currently getting a preview of what the processors will look like. future. It is really showing us how much we can greatly scale the power of our processors through the innovative use of micro combs, ”explained Dr Xu.
RMIT’s Professor Mitchell also added: “This technology is applicable to all forms of processing and communication – it will have a huge impact. In the long run, we hope to build fully integrated systems on a chip, significantly reducing costs and energy consumption. “” Convolutional neural networks have been central to the artificial intelligence revolution, but existing silicon technology is increasingly presenting a bottleneck in processing speed and energy efficiency, “said a key supporter of the research team, Professor Damien Hicks, from Swinburne and the Walter and Elizabeth Hall Institute.
This breakthrough shows how new optical technology makes such networks faster and more efficient and is a profound demonstration of the benefits of interdisciplinary thinking, in having the inspiration and courage to take an idea. from one field and use it to solve a fundamental problem in another, according to scientists. (ANI)
(This story has not been edited by Devdiscourse staff and is automatically generated from a syndicated feed.)
- A Swinburne University of Technology demonstrated that the world’s fastest and most powerful optical neuromorphic artificial intelligence (AI) processor is capable of operating faster than 10 trillion operations per second and capable of processing large-scale data.
- Scientists exemplify the world’s fastest optical neuromorphic processor for AI