Two CNS Faculty Awarded Funding: Program Accelerates Early-Career Faculty Research for Engineers and Computer Scientists

Six teams from the University of California San Diego Jacobs School of Engineering have been awarded funding through the School’s unique program designed to accelerate interdisciplinary research collaborations for early-career faculty.

The big idea is to empower early-career faculty to build interdisciplinary research collaborations to the point that they are competitive for multi-year research funding. The effort is funded by Irwin Jacobs and his late wife, Joan.

“This is a powerful program for our early-career faculty. I nickname it the two-two-and-two program. Two graduate students from two faculty labs are funded for two quarters in order to get a new cross-disciplinary research program up and running,” said Albert P. Pisano, Dean of the UC San Diego Jacobs School of Engineering and Special Adviser to the Chancellor. “It’s an effective mechanism for accelerating the research efforts of early-career faculty while also nurturing the collaborative culture of the Jacobs School among both our faculty and our graduate students.”

This year, that translates to a new cohort of twelve faculty and twelve graduate students, working together to build new research collaborations that will be competitive for external funding.

“I am deeply grateful to Irwin and Joan Jacobs for their vision, wisdom and generosity,” said Pisano. “It is a true joy to see what our early-career faculty and their graduate students can do with resources dedicated to building research bridges across labs.”

In particular, the program provides funding for graduate students from two different labs to begin new research collaborations. At least one of the two Jacobs School faculty must be an early-career professor. By funding research collaborations that link two different disciplines, the program provides exciting research experiences to graduate students.

The funded collaborations span a wide range of areas, often connecting disparate disciplines in unexpected ways that could open up new areas of research capable of helping to solve big challenges facing society. Though not by design, this year all six teams plan to make use of artificial intelligence and machine learning.

The newly funded projects include efforts to: create safe and reliable automated decision-making systems; invent a new generation of computer operating systems; better understand the genetics of heart disease; improve our ability to turn atmospheric carbon dioxide into useful industrial chemicals; develop load-bearing beams that are 3D printed from recycled plastics; and advance wireless communication systems.

“We had so many strong applications, and I am grateful to everyone who submitted proposals,” said Javier E. Garay, Associate Dean for Research and Professor of Mechanical and Aerospace Engineering at the UC San Diego Jacobs School of Engineering. “It is clear that our early-career faculty are extremely adept at seeing promising research connections across traditional domains. I am personally inspired by the creativity and technical rigor that is evident in each of these projects.”

Summaries of the six newly funded projects are below.

Retinomorphic Infrared Imaging System with Low-Power Edge-Inference Architectures
Mingu Kang, electrical engineering professor
Tina Ng, electrical engineering professor
For autonomous driving under low visibility, biomedical imaging and other applications, there is great interest in safe and reliable automated decision-making systems that rely on computer vision and function in real time at low power. One emerging approach relies on in-sensor computing, which places critical computing resources within the image sensor itself. A Jacobs School electrical engineering team is now working to create an in-sensor computing system which senses infrared light, rather than visible light. The project brings together electrical engineering professor Mingu Kang, an expert in CMOS ML architecture designs, and electrical engineering professor Tina Ng, an expert in new sensor devices including shortwave infrared semiconductors. The system they are developing combines in-sensor feature extraction with a sparsity-adaptive ML accelerator to generate feature-extracted images and enable real-time decision making while using very little power. This project aims to advance the design and processing knowledge necessary for enabling next-generation infrared imagers embedded with CMOS edge-computing.

Propelling CO2 Reduction: High-Entropy Alloy Design for Formic Acid Production Integrating Advanced First-Principle Calculations and Machine Learning
Wanlu Li, chemical and nano engineering professor
Yufei Ding, computer science and engineering professor
There is intense interest in developing more efficient and less expensive ways to remove carbon dioxide from the atmosphere by converting it to formic acid, which is a useful industrial compound. One of the most promising approaches to converting atmospheric carbon dioxide to formic acid uses high entropy alloys (HEAs) as catalysts. These catalysts, however, are incredibly complicated. A Jacobs School team is confronting this complexity by combining first-principle calculations and machine learning methods to design better high entropy alloys while creating a workflow that balances accuracy and efficiency. The team combines advanced first-principles domain knowledge from chemical and nano engineering professor Wanlu Li with machine learning expertise from computer science professor Yufei Ding. Bringing these two capabilities together, the team will go on to assess the scalability of their new-catalyst candidates while improving upon today’s theoretical methods for assessing high entropy alloys. This collaboration is poised to make important advances at the intersection of machine learning, materials science and the zero-carbon economy of tomorrow.

Operating Systems for the Age of Accelerators
Amy Ousterhout, computer science professor
Alex C. Snoeren, computer science professor
Today’s computer operating systems are not well equipped to handle the many fixed-function hardware accelerators that appear on recent generations of computer chips from both Intel and AMD. These hardware accelerators are designed to improve efficiency, lower power consumption and decrease operating costs despite the end of Moore’s Law and Dennard Scaling. But for hardware accelerators to be as useful as hoped, computer operating systems need to change. A new collaboration between two Jacobs School computer science faculty aims to address this issue by creating new operating systems that make the most of hardware accelerators. This collaboration leverages computer science professor Amy Ousterhout’s extensive prior work in user-level scheduling and computer science professor Alex Snoeren’s recent advances in application-informed memory management. Graduate students from their labs are now working together to develop a framework for managing fine-grained accelerators in userspace to make the most of accelerators for tasks ranging from compression and encryption to analytics, video encoding, ML training and inference.

Adaptive EdgeRIC for 6G Wireless Communication Systems
Jorge I. Poveda, electrical engineering professor
Dinesh Bharadia, electrical engineering professor
With the goal of enabling ultra-reliable, low-latency and high-bandwidth mobile communications, electrical engineers at the Jacobs School are working together to develop better ways to integrate control theory, reinforcement learning (RL) and 6G technologies. The electrical engineering team aims to leverage this approach to produce significant advancements in network management and optimization. These kinds of advances are critical for developing future higher-performance communications technologies that would be needed for a wide range of applications, including autonomous cyber-physical systems. The effort is a collaboration between two electrical engineering faculty at the Jacobs School: Jorge Poveda, an expert in nonlinear control theory and hybrid control; and Dinesh Bharadia, a specialist in wireless communication systems and networks. The team is leveraging a recently introduced monitoring and control microservice called EdgeRIC, created by the Bharadia lab and others. In the new project, graduate students from both labs will collaborate on developing a multi-time scale predictive control model; implementing reinforcement learning algorithms; and integrating their advances with 6G wireless systems.

Multiscale Mechanics and Fabrication of Additively Manufactured Recycled Thermoplastic Composites
Shabnam Semnani, structural engineering professor
Georgios Tsampras, structural engineering professor
Two structural engineering professors are collaborating to discover whether an emerging class of 3D-printed plastic materials is well-suited to serve as large load-bearing beams and other structural materials in constructing buildings, piers and other large structures. While these new materials have great potential in terms of sustainability and resilience, their structural and mechanical properties must be better understood before they can serve as load-bearing elements. To do this, graduate students from the two labs are working together to set up a system to fabricate, test and model a variety of different 3D-printed components made from this promising class of materials called recycled fiber-filled thermoplastic components. The researchers are focusing on how the processing parameters of 3D printing and the component size affect material properties. Structural engineering professor Georgios Tsampras’ contributes his expertise in creating automated robotic hybrid manufacturing systems that can produce large structural materials with complex shapes. Structural engineering professor Shabnam Semnani contributes her expertise in developing techniques for multiscale modeling of the mechanical behavior and damage in structural materials.

Harnessing Pangenomics for Complex Trait Association Studies
Yatish Turakhia, electrical engineering professor
Melissa Gymrek, computer science professor
Jacobs School researchers are taking a new approach to studying the relationship between genes and complex traits, such as heart disease and mental health conditions. Today’s methods for studying the genetic basis of conditions with complex traits are often limited by a reliance on a single reference genome. In contrast, the new approach considers the entire collection of genomes from the species. This approach is called pangenomics and aims to take genetic variation within the species into account. The new project is a collaboration between electrical engineering professor Yatish Turakhia, an expert in developing innovative data structures and file formats for pangenomics, and computer science professor Melissa Gymrek, a leader in linking genomic variations to complex traits. Their project is expected to provide new insights into the molecular basis of complex traits, shed light on their evolutionary and demographic histories and ultimately lead to more effective personalized medicine, public health strategies and genetic risk prediction.

Story by: Daniel Kane – dbkane@ucsd.edu