There’s a challenge in computer science – specifically, in face-tracking software that cannot read the faces of people with dark skin. Chinasa Okolo ’18, a computer science major and daughter of Nigerian immigrants, experienced this herself using the eye-tracking program WebGazer. WebGazer uses common webcams to track faces and predict eye-gaze locations of web visitors in real time so that it tracks where website visitors are focusing on the page and for how long.
For her senior project, Okolo decided to take a deeper look at WebGazer to help make it more comprehensive. “Fixing the bias in facial recognition and other artificial intelligence (AI) applications is important to making sure that people of all skin colors and backgrounds can use technologies that incorporate machine learning and artificial intelligence,” says Okolo.
WebGazer was developed by Pomona College Assistant Professor of Computer Science Alexandra Papoutsaki along with Professor Jeff Huan and his research group at Brown University.
“When I found out that the project she developed at Brown did not work for me – it wasn’t able to accurately track my face – I was immediately interested in working out and researching how I could solve the problem,” says Okolo, who has done a lot of reading about bias in artificial intelligence, and recently attended and presented research at the Black in AI workshop, which was part of the Conference on Neural Information Processing Systems in December. “Black in AI is a place for sharing ideas, fostering collaborations and discussing initiatives to increase the presence of Black people in the field of artificial intelligence.”
Okolo went to work quantifying how often the face tracking fails for people of color by gathering data sets, and Professor Papoutsaki welcomed the help. The professor notes that when she came to Pomona in fall 2017, her initial goal was to improve the accuracy of WebGazer’s gaze predictions. Then Okolo approached Papoutsaki and made the case for first fixing the facial detection algorithms that failed to work on her.
“It has been very rewarding to work with a student who has a personal investment in my research,” says Papoutsaki. “Chinasa embodies all the student characteristics that attracted me to Pomona in the first place.”
PATH TO POMONA
Okolo, who grew up in Kansas City, Missouri, says she benefited from local college-prep programs such as the Kauffman Scholars college access and scholarship program, and the Early College Academy, a partnership through which students enroll full-time at Metropolitan Community College–Penn Valley for the last two years of high school. And then, when she arrived to Pomona, she found the support she needed to succeed in STEM. QuestBridge and the Pomona Scholars of Math (PSM) were particularly helpful. “If I didn’t have them, I doubt I would have been as successful,” she says.
At Pomona, Okolo has honed her interest in artificial intelligence through a series of summer research experiences connecting computer science to medical and biology research.
The summer after her first year, Okolo spent 10 weeks at the University of Georgia on a project with their departments of computer science and infectious diseases, where she not only learned how to infect epithelial cells with listeria in order to track the rate of infection but she also learned how to code using Python, a programming language.
That sparked Okolo’s interest in identifying images, bio-imaging work and automating processes for analyzing images. Okolo spent the following two years doing research for Neuroscience Professor Elizabeth Glater, specifically, tracking the movement of the c-elegans roundworm using Matlab software.
The next summer, Okolo measured eye movements in humans and did computer programming and data analysis as a Howard Hughes Medical Institute Exceptional Research Opportunity Fellow at Columbia University.
“I’ve always been really passionate about biology and medicine, and I wanted to be a forensic pathologist, but I didn’t want to go to medical school,” says Okolo who was an avid fan of the television show Bones while in high school. After her internship and research experiences, she started to see the connections between biology and computer science.
This past summer, Okolo was at the Carnegie Mellon University where she helped revamp the user interface of a laboratory tool used to analyze cellular images obtained from microscopes.
“Being there helped me realize and strengthen my desire to go to graduate school,” says Okolo who has already been accepted to a number of top-tier Ph.D. programs in computer science, including Princeton and Cornell.
“My goals are to use artificial intelligence to aid in medicine and biology, in image analysis. The specific field is called computer vision. I’d like to use artificial intelligence and computer vision to improve modern image scanning techniques such as X-rays, ultrasound, CT scans and MRI, where computers would be better able to identify abnormalities that doctors may have missed.”
In her time at Pomona, she has also paid it forward. During her junior year, Okolo interned with the nonprofit Pomona Hope, revamping the organization’s technology education program and teaching computer coding to elementary school students.
Okolo’s advice to younger students is to find a balance between their interests. She stays involved in activities outside the lab that she’s also passionate about, including her role as manager of ReCoop, where she works to make the College more sustainable helping to recycle and salvage working items discarded by students.
And her commitment to inclusion isn’t limited to the realm of technology. Having a supportive environment for Black students on campus is critical for Okolo. As co-president of the Pan African Student Association (PASA), a 5-C student organization, she has played an integral role in coordinating the Sankofa Festival this past February and she’s currently helping organize the upcoming 6th Annual Black Hair Conference April 6.
“In PASA, we’re very good at engaging members of the Black community, we also try to be inclusive with other organizations, for any events, regarding diaspora and diversity in general,” and she adds with a smile, “We throw really good parties as well.”