Texas Engineers Play Critical Role in New NSF AI Institutes

Thursday, July 29, 2021

The National Science Foundation just announced 11 new artificial intelligence institutes across the nation, and researchers from the Cockrell School of Engineering at The University of Texas at Austin will play prominent roles in two of them.

The investment from NSF, $220 million in total, underscores the importance of AI in research today. The goals of these institutes include helping older adults lead more independent lives and improving the quality of their care; transforming AI into a more accessible “plug-and-play” technology; creating solutions to improve agriculture and food supply chains; enhancing adult online learning by introducing AI as a foundational element; and supporting underrepresented students in elementary to postdoctoral STEM education to improve equity and representation in AI research.

UT Austin was already among the top universities in the world for AI, and its involvement in these new institutes bolsters its strength in this emerging area. Last year, the university was selected by NSF to lead an AI institute focused on machine learning, the technology that drives AI systems, enabling them to acquire knowledge and make predictions in complex environments.

“Through our impressive representation in this year’s NSF AI Institute competition, Texas ECE has established itself as a leader not only in core machine learning but, most importantly, in translational and applied AI and machine learning,” said Diana Marculescu, chair of the Cockrell School's Department of Electrical and Computer Engineering. “Indeed, machine learning is impacting all aspects of how we live and interact with technology, and Texas ECE researchers are on the cutting edge of innovations that will make future AI applications possible, ubiquitous and safe.”

Here’s a look at the new AI institutes and how they will help move the evolving technology forward:


The AI Institute for Learning-Enabled Optimization at Scale, also known as TILOS, aims to “make impossible optimizations possible” by addressing the fundamental challenges of scale and complexity in several important areas, including semiconductor chip design, robotics and networks. The team is led by the University of California, San Diego, in collaboration with five other universities across the nation including UT Austin.

AI and machine learning are important tools for solving challenging optimization problems. In chip design, these advances would dramatically shorten the design schedule, while increasing quality, productivity and innovation in the process. In robotics, optimization breakthroughs would improve the way robots learn and interact with humans and other robots in many contexts, including search and rescue, autonomous transportation and inside warehouses. In communication networks, better optimization would lead to more capable and efficient power grids, mobile phones and ecosystems within the Internet of Everything.

Electrical and computer engineering professor David Pan co-leads the chips-systems team. Marculescu and Adam Klivans, professor in the Department of Computer Science and director of UT’s AI Institute for Foundations of Machine Learning, serve on the internal advisory board.

“Modern chip design is extremely complicated,” Pan said. “It requires algorithms, tools,and methodologies at different abstraction levels to deal with high complexity (e.g., billions of transistors) and nanometer-dimension transistors and wires, under many design objectives and constraints such as performance, power, manufacturability, security and more. Chip design has long inspired optimization innovations, from simulated annealing to randomized rounding and spectral embedding.”

Through TILOS, Pan and the chips-systems team will work closely with the AI/optimization foundations team to tackle key challenges from layout to verification. They will explore new paradigms and algorithms to perform electronic design automation.

The researchers will work with industry partners including AMD, Cadence, Google, IBM, Intel, NVIDIA, Samsung, Synopsys and others, on real-world problems and technology transfer. The research agenda is accompanied by plans for workforce development and broadening participation at all academic levels, from middle school to advanced research levels, including community outreach efforts to promote AI.


The AI Institute for Future Edge Networks and Distributed Intelligence, called AI-EDGE, is led by The Ohio State University, and it aims to design future generations of wireless edge networks that are highly efficient, reliable, robust and secure. New AI tools and techniques will be developed to ensure that these networks are self-healing and self-optimized.

This team includes electrical and computer engineering professors Sanjay Shakkottai, Constantine Caramanis and assistant professor Aryan Mokhtari. Their expertise is in bridging wireless networks and machine learning. Shakkottai is leading one of the research thrusts of the project, AI-based network resource allocation. And the UT Austin researchers will also work on other key project areas, including multi-agent network control, network aware AI operation and network operations for distributed AI.                

"This Institute draws on long-standing strengths within UT in wireless networks and AI/ML,” Shakkottai said. “We will collaborate with researchers in the Wireless Networking and Communications Group — the home of 6G wireless research at UT — the UT Austin Machine Learning Lab and the NSF AI Institute for Foundations of Machine Learning. We are thrilled to have this opportunity to collaborate and develop critical technologies across these disciplines.”

The team is focused on edge networks — a trend in the technology world where data storage and processing occur closer to devices and users — because the majority of growth is expected to come with wireless devices, services and applications at the edge rather than the traditional network core. These edge networks will encompass mobile and stationary end devices, wireless and wired access, and computing and data servers.

Collaboration over these adaptive networks will help solve long-standing distributed AI challenges, making AI more efficient, interactive and privacy-preserving for applications in sectors such as intelligent transportation, remote health care, distributed robotics and smart aerospace. It will create a research, education, knowledge transfer and workforce development environment that will help establish U.S. leadership in next-generation edge networks and distributed AI for many decades to come.

The highly collaborative project includes 30 scientists and engineers from 11 institutions, three U.S. Department of Defense research labs and four global companies.