Waheed U. Bajwa, Associate Professor of Electrical and Computer Engineering at Rutgers University, New Brunswick
Waheed U. Bajwa is a visiting fellow in the Center for Statistics and Machine Learning at Princeton University for the 2021-2022 academic year.
Bajwa received a bachelor’s degree in electrical engineering from the National University of Sciences and Technology, Pakistan, in 2001. He then went on to earn a master’s degree in electrical engineering from the University of Wisconsin-Madison in 2005 and his Ph.D. in 2009 from the same institution.
After earning his doctoral degree, he assumed a postdoctoral research position at Princeton University’s Applied and Computational Mathematics Program from 2009 to 2010. He was also a research scientist at Duke University’s Electrical and Computer Engineering Department from 2010 to 2011.
In 2011, he joined Rutgers University, New Brunswick, where he is currently an associate professor in the Department of Electrical and Computer Engineering and an associate member of the graduate faculty of the Department of Statistics.
During his academic career, Bajwa has received numerous awards for his research and teaching, including Rutgers University’s Warren I. Susman Award for Excellence in Teaching, the Army Research Office Young Investigator Award, the National Science Foundation CAREER, and the Cancer Institute of New Jersey’s Gallo Award for Scientific Excellence, among many others.
Bajwa has held several co-chair positions for conferences, symposiums, and workshops and serves as an editor for several scientific journals.
Bajwa’s research is at the intersection of statistics, machine learning and signal processing. The latter discipline is an engineering subfield that concerns itself with analyzing or processing images, sound, and other data.
“I try to build algorithms for machine learning and statistics and signal processing with a particular focus on applications in which data is gathered from sensors,” he said. “I’m particularly interested in situations when this data is obtained from engineering or natural systems, such as climate data, microscopy data, or fMRI data.”
Bajwa and his lab seek to understand the mathematical basis of problems. Then he develops “theoretically optimal, computationally efficient, and algorithmically robust solutions.”
“We don’t collect the data, but we work with collaborators who generate or collect the data,” he said. “For example, we have a collaborator in particle physics and one in biological imaging.”
He and his lab work on data of four types. The first consists of high-rate data, such as Internet-of-Things applications or physics. The data in these domains need to be processed quickly, said Bajwa. The second type is large-scale data that tends to be distributed or decentralized. The third type is data that has multidimensional aspects such as colored video, which has distinct dimensions for space, color and time.
“Many problems involve multidimensional data sets, but we tend to ignore the multidimensionality. Traditionally we squish it into one dimension, which is not the best approach to take in both a computational or performance sense,” said Bajwa.
The fourth type of data is either corrupted maliciously or inaccurate due to limitations of the sensing systems, he said.
His recent publications include work on imaging, such as “Computational endoscopy – a framework for improving spatial resolution in fiber bundle imaging,” published in the journal Optic Letters in 2019, and “Sample Complexity Bounds for Dictionary Learning from Vector- and Tensor-valued Data,” published in the book “Information-Theoretic Methods in Data Science" by Cambridge University Press in 2021.
Bajwa sees his time on the Princeton campus as a chance to get back into pure research and to start or continue projects with collaborators on and off-campus. He’s working on particle physics problems and issues with non-Euclidean irregular data, such as how can researchers build algorithms for that scenario? That project is in its infancy, said Bajwa, who is exploring the problem with a few possible collaborators in CSML.
“CSML provides a vibrant environment for a lot of people who are doing interesting stuff within the realm of machine learning,” he said. “I am here to exchange ideas and see what is happening on campus. The fundamental goal is the advancement of science.”