AsianScientist (Dec. 29, 2016) – Researchers at Tohoku University have, for the first time, used spintronics to develop artificial intelligence (AI). Their findings, published in Applied Physics Express, have the potential to dramatically reduce power consumption, a major limiting factor for AI.
Existing AI applications use the conventional framework of semiconductor-based integrated circuit technology. However, this lacks the compactness and low-power feature of the human brain. Spintronics, which uses the spin rather than the charge of the electron to store information, is one promising way to overcome the size and power constraints currently faced by AI researchers.
Using a recently-developed spintronic device based on micro-scale magnetic materials, a team of researchers from Tohoku University led by Professor Hideo Ohno has developed the first ever spintronics-based artificial neural network. Unlike conventional magnetic devices, the spintronics-based device is capable of memorizing arbitrary values between 0 and 1 in an analogue manner.
Using the developed network, the researchers examined an associative memory operation, a function not readily executed by conventional computers. Through the multiple trials, they confirmed that the spintronic devices have a learning ability and that the artificial neural network can successfully associate memorized patterns from noisy inputs, just like the human brain can.
This proof-of-concept demonstration is expected lead to new AI technologies that are compact, fast and yet low powered. These features should enable AI to be used in a broad range of societal applications such as image/voice recognition, wearable terminals, sensor networks and nursing-care robots, the researchers said.
The article can be found at: Borders et al. (2016) Analogue Spin–orbit Torque Device for Artificial-neural-network-based associative Memory Operation.
———
Source: Tohoku University; Photo: Shutterstock.
Disclaimer: This article does not necessarily reflect the views of AsianScientist or its staff.
