The University of Tennessee Research Foundation is proud to support researchers in emerging and developing research fields, including neuromorphic computing, a computing system design modeled after the human brain.
The TENNLab Neuromorphic Computing Research Group is made up of faculty, postdocs, graduate students, and undergraduates who are developing hardware and software to aid in the development of revolutionary computing paradigms. Together, they bring expertise in computer science, computer engineering, electrical engineering, neuroscience, material science and physics.
Currently, four faculty members from UTK’s Department of Electrical Engineering and Computer Science lead the research group: Professor and Associate Department Head Garrett Rose, Professor James Plank, Assistant Professor Catherine Schuman, and Assistant Professor Ahmedullah Aziz. Professors Emeriti and original TENNLab founders also include Mark Dean and Doug Birdwell.
What’s great about neuromorphic computing is the field’s interdisciplinary nature. It’s basically impossible to do research in this field in isolation,” said Schuman. “There also aren’t a lot of people in the field, so with almost everything we’re doing, we’re first people to try and see what will work. Exploring new ground is really exciting.”
Since its foundation in 2014, the group has worked with the University of Tennessee Research Foundation to file numerous invention disclosures, resulting in seven issued patents and four pending patent applications. TENNLab is one step closer to building intelligent, extremely low-powered systems that can be leveraged in real-world applications.
TENNLab has a broad portfolio of technologies that are paving the way for future discoveries in neuromorphic computing,” said UTRF Vice President Kusum Rathore. “With an increasing need for alternative approaches to traditional computing architecture, TENNLab is at the forefront of computing innovation.”
A unique challenge of neuromorphic computing is understanding how to leverage the unique capabilities of these systems. TENNLab has created three novel frameworks, NIDA, DANNA, and mrDANNA, for dynamic, spiking neural networks – the basis of work for the group’s seven issued patents and the foundation of Schuman’s dissertation work and inspiration for creating TENNLab. Further inventions from TENNLab are related to memristive non-volatile memory, spiking neural network hardware, spiking neural network communication class optimization, and cryogenic neuromorphic hardware.
The innovations developed by TENNLab are more than just patents and encompass nearly all facets of neuromorphic computing. Examples of these innovations include software and know-how related to hardware implementation, theoretical models, programming techniques and applications, makers kits, software frameworks and much more. To learn more, click here.
One of TENNLab’s strengths is its collaborative approach, which Schuman calls “co-design.” Even with their different projects and specializations, they must constantly communicate to ensure their work aligns with the larger picture.
You might solve one issue at the device level, but then that creates software or algorithm-level challenges,” said Aziz. “Working together makes it easier for us to iterate and find better solutions.”
With increased industry interest and investment in artificial intelligence lately, Rose believes the research community is more interested in pushing the envelope and exploring new possibilities with neuromorphic computing and other bio-inspired systems.
When I started in the field, we were in what is referred to as an AI winter, but people have always explored and been interested in AI,” said Rose. “Computing has its roots in trying to understand how the brain works.”
Rose, Plank, Schuman, and Aziz are thankful for UTRF’s support and the many collaborators, postdocs, and students they have partnered with that make their current success possible – from students developing applications for autonomous video game playing to funding opportunities from the U.S. Air Force.
Everyone wants to build a brain. It is the easiest recruiting tool because people are fascinated by the brain. I can tell you when I researched storage systems, I did not have students flocking to me like we do now,” joked Plank.
For more information on TENNLab’s Neuromorphic Computing & Memristive Technologies, click here.