2023-07-22 22:10:00
The technological explosion of recent years has generated an exponential growth in global computing needs. The scientific community is striving to improve current processing capabilities while developing new computational methods to meet this critical need.
The research work carried out by the group of Jean Anne Incorvia, professor at the Cockrell School of Engineering’s Chandra Family of Electrical and Computer Engineering, are a perfect illustration of this approach.
Two new publications from the research group aim to make significant contributions to both current and future scientific needs. They offer improvements for current semiconductor technology while providing more nimble building blocks for the next generation of computers that mimic the workings of the human brain.
“We’re on the cusp of a new generation of computers that mimic the way our brains work, a massive research undertaking“said Jean Anne Incorvia. “At the same time, the computing techniques we use today will not disappear, so it is important to continue to improve and innovate the devices that power our current technology..”
The logic problem of transistors and circuits
A new study published in ACS Nano deals with transistors and circuits. Components called logic gates that interpret digital signals are built into the chips. Logic gates are transistors that can usually conduct either electrons or holes (which occur when electrons move inside atoms), but not both. In this study, the researchers connected logic gates capable of conducting both electrons and holes.
They demonstrate that this realization reduces the number of transistors needed in a circuit. This means that more transistors could be packed into the same space, making the device more efficient and powerful, or the space saved could be used to reduce the size of the device. They also demonstrate a new circuit that specifically exploits the behavior of the transistor.
Ultra-thin materials for ambi-polar transistors
Transistors are made of ultra-thin two-dimensional materials, which have this property “ambi-polar” which allows them to conduct both holes and electrons. However, they weren’t doing it very well on their own. Improving this capability is a major component of this document, and through their device engineering, they show significant XOR, NOR, and NAND gates without the need for any devices other than ambi-polar transistors. These circuits are the building blocks of larger circuits.
“When we think about the future of computing, if we can harness the natural behavior of these 2D materials and scale them, we could halve the number of transistors we need in our circuits.“, commented Jean Anne Incorvia.
Artificial neurons resist noise
A second document recently published in Applied Physics Letters looks at the next generation of computers, those that think more like the human brain. These neuromorphic devices are better than traditional computers for AI tasks such as image interpretation and language processing.
In this new paper, the researchers created a new type of artificial neuron – which in the human brain is responsible for sending information between brain cells – using magnetic materials. What sets these neural devices apart is the chaotic nature of their reactions to electrical impulses.
They outperformed other artificial neurons as part of neural networks to interpret images, especially when the data to be interpreted was noisy. The devices performed better than other artificial neurons in identifying fuzzy shoe images, and the gap widened as the images became fuzzier.
The implications for on-board computing
These neurons could have a significant impact for “edge computing” uses, where devices need to be smaller, consume less power, and be located away from a centralized computing source like a cloud server. They are also radiation resistant.
One of the first applications of this technology could be in space, where silicon chips struggle to withstand high levels of radiation. The ability to handle radiation as well as messy data could make these neurons ideal for future space technology.
Synthetic
Scientific advances in computing are increasingly promising, offering both improvements for existing technology and new perspectives for the next generation of computing systems. Recent research carried out by the group of Jean Anne Incorvia is a remarkable example of this evolution, by proposing more efficient transistors and developing new types of artificial neurons.
For a better understanding
What is an ambi-polar transistor?
An ambi-polar transistor is a device that can conduct both electrons and holes, a property usually absent from conventional transistors. This capability can allow more efficient and compact circuits to be constructed.
What is an artificial neuron?
An artificial neuron is an electronic device designed to mimic the function of neurons in the human brain. They are commonly used in artificial neural networks for information processing.
What is on-board computing?
On-board computing is an approach that aims to process data closer to the source where it is generated, rather than sending it to a remote centralized server. This can reduce latency and improve the performance of computer systems.
What is Radiation Resistance?
Radiation resistance refers to the ability of an electronic device to function normally in environments exposed to high levels of radiation, such as outer space. Radiation resistant devices are essential for space technology and some military applications.
Applied Physics Letters research was supported by grants from the National Science Foundation. The Incorvia team on this project is made up of Thomas Leonard, Samuel Liu and Harrison Jin, all electrical and computer engineers.
[ Rédaction ]
1690075823
#future #computing #ambipolar #transistors #artificial #neurons