A Ph.D. for Tacoma
Standing up a brand-new research doctoral program in computer science is not quite rocket science, but it comes close.
Especially when the context is a university campus that is barely more than 30 years old, with only one other doctoral program (the Ed.D. in educational leadership).
“Our original vision was to enroll 1-3 students each academic year,” said Joel Larson, director of operations in the School of Engineering & Technology, home to the new degree program. “We started the program in autumn 2021.”
“We currently have 20 active Ph.D. students. We were expecting to have between three and nine students at this stage.”
Celebrating the Milestone
The first two students to complete their Ph.D.s with the School of Engineering & Technology may defend their dissertations as early as this summer. Abdulrahman Salama’s work, supervised by Dr. Mohamed Ali, uses neural networks to identify inconsistencies in commercial mapping services like Google Maps, Bing Maps and OpenStreetMap. Adel Sabour, also supervised by Dr. Ali, seeks to bring Natural Language Processing systems, heretofore developed mainly using the English language, to the Arabic language.
Runaway Success
What accounts for the program’s runaway success?
“First, we have terrific faculty,” said the school’s dean, Raj Katti, noting that the reputation and the research focus of the faculty is what attracts students who want to work with them. “Second, we have strong support in the community,” he added, citing tech employers throughout the Puget Sound region who are calling on higher education institutions to help them fill their research labs and expand the tech workforce. “And, third, we have some very generous friends who want to see this happen at UW Tacoma.”
On that latter point, Katti is referring specifically to former UW Tacoma Chancellor Vicky Carwein and her husband, retired engineer Bill Andrews, who have set up an endowment to support SET’s doctoral program.
“Vicky and Bill’s investment in UW Tacoma and the South Sound community is an incredibly visionary event in our history,” said Jill Purdy, a founding faculty member in the Milgard School of Business who served as the campus’s Executive Vice Chancellor for Academic Affairs when the SET Ph.D. program was being developed.
“At the doctoral level, incoming students are typically recruited to work with particular faculty, with offers of funding to serve as teaching and research assistants,” said Purdy. The funding is often drawn from a combination of endowments, research grants and teaching assistant salaries. In some cases, an employer or government entity covers the cost.
Attracting talent is pivotal to building and maintaining a Ph.D. program’s reputation. Financial support can be a recruiting tool, and the lack of it a barrier for some students. Supports like the Carwein-Andrews endowment and partnerships with companies and other organizations play a crucial role, allowing SET to compete on the global stage to attract gifted students and researchers.
Use-inspired Innovation
“The SET Ph.D. is distinctive because of its focus on use-inspired innovation and interdisciplinarity,” said Purdy. “It encourages students to draw upon multiple disciplines within engineering and computer science to answer practical questions and solve real-world problems.”
Purdy and Katti both cite the role the Ph.D. will play in creating technology leaders in the South Sound. “The program will spawn new enterprises in fields like cryptography, machine learning and quantum computing,” said Purdy. “It positions UW Tacoma as a hub for technology innovation that can support community and economic development throughout the South Sound.”
Examples of such use-inspired innovation are coming fast and furious. Two quick examples: one about diagnosing leukemia, and another about reining in the privacy-violating excesses of artificial intelligence.
Harnessing Data for Cancer Diagnosis
Shishir Reddy came to UW Tacoma from California. He got his undergraduate degree at UC Irvine. An internship with Dr. Cecilia Yeung at Fred Hutch Cancer Center brought him to Seattle. Through Cecilia, he met SET Professor Dr. Ka Yee Yeung (no relation).
Reddy took advantage of a “twofer” admissions option, applying and being admitted to SET on a masters-to-doctorate path. He completed his M.S. in computer science and systems in 2022 and is now working on the Ph.D. The Carwein-Andrews endowment helped support his master’s-level coursework and he also received support via a UW Graduate School Top Scholar Award.
“I am working on using machine learning to improve leukemia diagnosis,” said Reddy. “On a high level, we take a blood sample, we extract DNA from the sample, we align the DNA with a map of the human genome, and then we look for mutations or variants.”
The work relies on the fact that there are some genetic abnormalities in cells that are strong indicators of different forms of leukemia. Techniques to see these abnormalities were first developed in 1960, and advances have continually shortened the time the examination takes. Reddy said the artificial-intelligence-based system he is collaborating on will shorten the diagnostic time, with high accuracy, down to at most a few hours from what previously took days or a week.
Privacy-Enhancing Technologies
The Carwein/Andrews endowment is supporting Sikha Pentyala and Steven Golob as they pursue their Ph.Ds., studying privacy-protecting machine learning (PPML).
The two are working with Professor Martine De Cock and Associate Professor Anderson Nascimento, who have been doing ground-breaking research in PPML for several years (see “Protecting Privacy by Sharing Secrets”).
Sikha Pentyala, an Indian immigrant, is on a mission to change the field of artificial intelligence, making it trustworthy and ethical. Trustworthy AI aims to build machine-learning and intelligent systems that respect users’ privacy, are safe to use and robust against adversarial manipulations of those systems. Trustworthy AI also should not discriminate against specific subsets of its users, a property known as “fairness.”
Pentyala is a trailblazer in trustworthy and fair AI systems. She has already published her research in leading AI journals, started a research collaboration with Montreal’s Mila, a global leader in AI research, and developed an open-source library for privacy-preserving fairness auditing of machine-learning models.
Moreover, as the mom of a four-year-old daughter, Pentyala is breaking down barriers for women in a field where they are underrepresented, serving as an inspiration to others who face similar obstacles. She recently learned she has been awarded a prestigious fellowship from JP Morgan Chase, one of the world's largest banks, joining other fellows from institutions like Cornell, Carnegie Mellon, Oxford, Princeton, Stanford and UC Berkeley.
Steven Golob grew up in Puyallup. “I was always good at math,” said Golob. “I wasn’t really a computer person. Now I am, and I’m very good at the logical, analytical aspect. I like to focus on theory.” Golob eventually graduated with honors from UW Tacoma with a degree in computer science and joined Boeing.
When he learned about the SET Ph.D. program, he decided to return to school, and brought with him a prestigious Computer and Information Science and Engineering Graduate Fellowship from the National Science Foundation. That fellowship was awarded to only 15 other students in 2021, who are now enrolled at institutions across the U.S., such as USC, Stanford, Yale, NYU and UC Berkeley.
Pentyala and Golob are working with De Cock and Nascimento (and others) on privacy-protecting methods that, to a layperson, seem like magic. The dawning age of artificial intelligence requires things like chatbots to be “trained” on massive quantities of data, some of which may contain personally-identifiable information. Using techniques based on cryptography and federated “learning,” the UW Tacoma team is creating tools that allow AIs to benefit from the content of data without needing to know the details of that content.
“Individuals are not consulted when their data is used to train machine learning systems,” said Pentyala. “We are developing methods to protect the privacy of users that will enable them to benefit from machine-learning-based systems without compromising their privacy.”
Domains of Knowledge
What these teams of faculty and students develop should be usable with many kinds of machine-learning systems across many domains of knowledge. For example, De Cock and Nascimento earlier applied their PPML techniques to systems being trained on social media posts. Most recently, one of their teams won second prize in an international competition involving financial crime prevention, and another of their teams achieved a top ranking in a NASA competition to improve the U.S. commercial aircraft flight network.