Vishal Saxena: Micron Endowed Professor
Designing Computer Chips Inspired by the Brain, Made Faster by Manipulating Light
Imagining a world in which artificial intelligence rules the day isn’t hard for Vishal Saxena — though he admits that such a sci-fi future is still 100 years away.
Before autonomous drones inundate low-altitude airspace, however — transporting people to and from their destinations, delivering goods, tracking reintroduced wildlife and endangered species, and monitoring disease in large-scale farming operations — certain problems with artificial intelligence and the memory chips that serve them need to be solved.
Machines need to become better at reasoning and perception — a subset of AI known as “deep learning,” learn lessons autonomously so they can reprogram without human supervision, and operate on less power to reduce their dependence on cloud infrastructure, which has a significant carbon footprint.
Saxena, who became the Micron Endowed Professor in Microelectronics in U of I’s College of Engineering in 2016, is working on solutions to these problems. In fact, the professor and award-winning researcher is playing a crucial role in developing some of the most innovative technology on today’s semiconductor market.
Saxena is developing neural-inspired chips — hardware that could be used in machines with artificial intelligence — that emulate the brain in its energy efficiency and capabilities.
Computing systems with deep learning capabilities are already driving the business industry. They can recognize speech and be used for automatic translation services, and they can perceive images for safety features in cars to alert drivers to large animals in the roadway. The systems require expansive data centers with clusters of servers that consume considerable amounts of energy, though.
Since the brain surpasses computers in its ability to reason and perceive images, and because it’s very low power — “even more low power than a 60-watt incandescent light bulb,” Saxena said, the microelectronics professor is striving to better understand the human processing system in order to develop his neuromorphic chip.
Since arriving at U of I, Saxena was able to bring together a group of engineering faculty from various disciplines to form the Neural Engineering and Brain-inspired Systems (NeuBIS) group. Their goal is to use their collaborative knowledge of the brain, from an engineering standpoint, to develop new computing and memory systems that have strong synergy with Micron.
As part of the endowment, Micron has enabled Saxena to hire additional student researchers and expand course offerings, the result of which will be a well-educated labor force in the field of microelectronics. Saxena also purchased state-of-the art equipment that allows him to test high-speed circuits in-house, rather than send students to industry labs in Palo Alto.
Other technology that Saxena and his team are developing include next-generation hybrid electronic-photonic chips. After working for a startup tech company in Pennsylvania, he found revolutionary ways to harness light onto circuits, rather than solely using electrons.
In one application, the chips reduce energy consumption and increase data storage capacity in internet cloud data centers. In today’s internet-centric society, with cloud-based services such as Google Docs, Dropbox and Netflix making big data more accessible, the carbon footprint of data centers is massive. Saxena’s research directly addresses that problem. In 2014, the idea won him five years of funding from the National Science Foundation CAREER Award.
These hybrid chips also have applications in wireless devices for the Department of Defense. They allow military personnel to consolidate their devices, such as cell phones and GPS systems that require different frequencies, into one mobile receiver called a flexible radio. For this project, in 2015, Saxena won three years of funding from the Air Force Office of Scientific Research Young Investigator’s Award.
Once complete, Saxena hopes these memory chips will be of manufacturing interest to the semiconductor industry.
The Lucknow, India native has come a long way since the time in 1991 when he built his first video game on a BBC microcomputer with only 16 kilobytes of memory. He hopes his imagination serves Micron well, too.
“We need to make progress,” Saxena said. “For the sake of scientific curiosity and from an economic point of view, we want to come up with better and more interactive computing products that will drive the future of the semiconductor industry. The goal is one — U of I is building a center of excellence in electronic circuits synergistically with Micron.”
Article by Kate Keenan, College of Art and Architecture