Comp Sci Surge

As the world becomes increasingly data driven, the field of computer science has exploded with students—and there simply aren’t enough professors to teach them. At Haverford, as at many other colleges across the country, overcrowded classes have resulted in lotteries and course caps, leaving some computer science majors scrambling to complete requirements.

“This is part of a broader national trend,” said Associate Professor of Computer Science Sorelle Friedler. “We have seen it here at Haverford even beyond the national trends. There’s been a lot of interest as there’s been more of a public understanding of the importance that tech companies play in democracy, for example, and just in our everyday lives.”

The issue at Haverford has improved thanks to the hiring of two tenure-track faculty members: Assistant Professor of Computer Science Sara Mathieson, who specializes in computational biology and began teaching in fall 2019, and Assistant Professor of Computer Science Alvin Grissom, who focuses on computational linguistics and is teaching his first classes this semester. The addition of Visiting Instructor of Computer Science Rajesh Kumar, who researches human-centered computing, biometrics and security, and applied machine intelligence, has further filled out the computer science faculty ranks—an expansion that will give students more options for exciting electives and help ensure that they will find seats in core courses.

The kind of strains on computer science resources that Haverford has experienced have become common. In a 2015 survey, the Computing Research Association showed that the number of undergraduate computer science majors in the United States was higher than at any other time, including during the dot-com boom of the late 1990s. The CRA also reported that over 60 percent of responding academic institutions more than doubled their computer science enrollment between 2009 and 2014. 

The numbers are similar at Haverford. In 2010, the College had just one computer science major, but in the spring of 2020, there were 29. The number of declared majors expected to graduate in 2021 and 2022 are 33 and 37, respectively. 

The department was forced to stop offering a minor in computer science last year in order to accommodate majors, but that didn’t entirely alleviate the problem. Classes had swelled to 30, sometimes 40 students. (And even those numbers, notes Mathieson, don’t represent the full scope of campus interest, as lotteries and caps held down class sizes that could have ballooned to as many as 80 students, according to preregistration estimates.)

“I was surprised by all the lotteries because of how popular computer science is,” said Alison Rosenman ’20, who graduated with a computer science degree in the spring and is now working as a technology rotational associate at the Capital Group in Irvine, Calif. “Even as a senior, I had to worry about course caps and lotteries. I worried that the lack of staffing decreased the rigor of the major.”

Professors had similar concerns: They wanted to make sure the department had the resources to offer a true Haverford education in their discipline.

But now, with two new faculty and a visiting professor on campus, the Department of Computer Science can better accommodate its students—both in terms of class size and elective offerings.

“This is really strengthening our breadth as a department,” said Professor and Chair of Computer Science Dave Wonnacott. “When we were really small, we focused on the classical field of computer science for its own sake—theoretical foundations and how systems work. Now we’re able to make richer connections to relevant purposes of computer science.” 

Haverford also has a new computer lab with 40 Linux machines that have graphics processing units, which are important for teaching new “Machine Learning” and “Computational Linguistics” courses. (And thanks to a clever installation by the College’s Instructional and Information Technology Services, the monitors and keyboards can switch from Linux to Windows by flipping a single switch.) The College also allocated funds for new research space and classroom renovations to accommodate the growing number of students.

Working through biological problems using computer science 

Computer scientist Sara Mathieson adds to Haverford’s existing interdisciplinary research with her focus on how biological problems drive innovation in algorithm development and how computational approaches can uncover new biology.

“The expansion of the department will allow me to teach a broader range in my field,” Mathieson said. “We’ll all be able to put more attention on our specialties, which is good for the students and advances our research, too.”

One of her current research projects aims to develop algorithms to reconstruct the genomes of ancient individuals based on present-day genomes. With these reconstructions, specifically in families with high rates of heritable disease, biologists will have new clues about what genes are responsible for disease risk. For her project, Mathieson is studying bipolar inheritance patterns in an Old Order Amish population from Lancaster, Pa.

“The results should help us better understand the genetic component of bipolar disorder, and the algorithms we’re developing could be used in other reconstruction problems where we don’t have direct access to the necessary DNA,” explained Mathieson, who has Haverford students working in her lab and collaborating with researchers at the University of Pennsylvania. “This same type of analysis could be applied to any other heritable disease in an extended family.”

Mathieson is teaching “Machine Learning” this fall, and she’s partnering with Assistant Professor of Biology Eric Miller to develop a “Computational Superlab” for the spring semester. While they haven’t sorted out all the details yet, it’s a course that will be greatly enhanced by the collaboration made possible by an expanded Department of Computer Science. The interdepartmental nature of the course illustrates the need for—and usefulness of—a level of computer science competency across many different disciplines. 

“I want to see computer science treated as a basic skill, like writing and math,” Mathieson said. “I want everyone, even non-majors, to be able to take computer science and see it for the interdisciplinary tool and subject that it is. That’s what we’re trying to do at Haverford.”

Friedler agrees: “Computational literacy is important in order to be an educated member of society."

Studying language through a social justice lens

Assistant Professor of Computer Science Alvin Grissom is excited to bring his expertise in computational linguistics to a liberal arts college. And students are just as eager to have that specialization in the classroom. 

“His research fits really well with what I’m interested in,” said Blien Habtu ’21, who is majoring in both computer science and linguistics. “I’m hoping this [department expansion] will give our professors a chance to offer more electives in their own research areas.”

There are two general ways computational linguists can focus on their discipline. One is to build tools that use language as a primary component, such as a Google search or an app that translates text from one language to another. When building a tool that must recognize speech in order to perform its function, such as Siri, a computational linguist must consider variables such as dialects and accents so the tool can understand all users’ speech. 

Another way to apply the discipline is to use computational tools to study language. For example, in one of Grissom’s research projects, he’s studying the language that commentators use to talk about American football. Grissom and his collaborators have analyzed the fast pace of football commentary and confirmed his hypothesis: that players of different races are discussed in different terms. This research has scientifically identified a phenomenon that may play a role in discussions of social justice. 

“I’ll incorporate these broad ethical concerns into my classes,” said Grissom, adding that the book Race After Technology: Abolitionist Tools for the New Jim Code by Ruha Benjamin is required reading in his “Computational Linguistics” course this semester. “This book reframes the technology that people like me develop in a social justice lens. We’ll talk about the implications for what the students are learning in the classroom.”

Those implications may involve issues where machine learning intersects with race—for example, when facial recognition systems misclassify nonwhite faces more often than white faces.

Wonnacott offers another example: Machine learning techniques can create a “race-blind” algorithm that leaves out racial properties of the collected data. However, if the algorithm includes information about income and zip code, those paired data sets can provide racist results unintentionally.

“We can then use other computational techniques to recover racial bias, so you can at least identify if it has picked up racist behavior,” Wonnacott said. “But when you’ve done that, what counts as fair and just in that context?”
    

Outsmarting smart technology

Visiting professor Kumar will continue his research on behavioral biometrics. Physical biometrics—such as facial recognition, fingerprints, or retina scans—can be used to identify an individual for many purposes, including security. Similarly, behavioral biometrics, such as walking patterns or other movements, can also be used to authenticate identity and will likely become more common in the future. Kumar is researching the ways such technology could be tricked—and security compromised.

“These movements are unique to you and form a signature,” explained Kumar, who is teaching “Introduction to Computer Science and Data Structures” and “Computer Security: Attacks and Defenses” this semester. “For example, imagine you have a smart car that can automatically unlock its doors based on the hand movement or walking pattern you produce when you walk toward the car. I’m studying to find out if someone else can walk like you, if someone mimics your pattern, will the system be fooled?”

Kumar also emphasizes the need for a more diverse field of computer scientists to address problems in developing security tools, such as facial recognition systems.

“We’re relying more on computer-based processes, which must be unbiased,” Kumar said. “We have to think about diversifying the teams, the people who are developing and implementing the software. When teams are diversified, they will be more likely to anticipate potential biases.”
 

Expanding, enriching, and still personal

With these faculty additions, Haverford will be able to continue offering the individual attention made possible by smaller classes. 

“One of my favorite things about Haverford is its small size, making it easier for students to maximize the amount of time they can collaborate with professors,” Habtu said. “Almost the entire department knows me by name and are willing to sit with me to chat about my academic goals.”

The department is pleased that Mathieson, Grissom, and Kumar will add new opportunities for interdisciplinary collaborations while also teaching students who choose to follow a traditional path that there are broad applications for the field of computer science. And students are eager to embrace the interdepartmental collaborations these professors’ research will make possible.  

“Being at a liberal arts school allows people to understand both the computational side of it and the more applied side of it in real in-depth ways,” Friedler said. 

“[The new professors’] expertise in machine learning, particularly with biology data and linguistics, is essential in both industry and academia,” said Rosenman. “I appreciate Haverford’s realization that computer science does not always equal coding.”

Broadening perspectives 

STEM fields historically have had little diversity, but the number of women and nonwhite students in the field is growing, nationally and at Haverford. For the 2019–20 academic year, nearly 60 percent of computer science students at Haverford were nonwhite, and 21 percent were women. 

As Haverford adds a woman and two people of color to its computer science faculty, students and professors alike recognize the advantages these new faces will bring. 

“Some students haven’t seen themselves as computer scientists or haven’t been encouraged to follow that path,” Sara Mathieson said. “It can be powerful to see those people in positions of authority and influence. If students find faculty that they identify with and resonate with, that will serve our majors as they think about themselves as the next generation of scientists.”

“I want every student to be successful, and the lack of representation can have a psychological effect on some students,” said Alvin Grissom, who is Black. “One way of addressing that problem is to have more Black faculty in computer science. Having Black mentors will increase the success rates of Black students. To the extent that my presence can give them an example of what’s possible, that’s great.”

“International students’ concerns and questions may get overlooked when the faculty is all domestic,” adds Rajesh Kumar, who is from India. “I bring a different experience and unique perspective about what we can do to support our international students.” 

“It’s powerful to see people [who look] like you leading and teaching,” said Blien Habtu ’21, who is Black. “I can imagine a first-year, or someone who is still trying to figure out exactly where they fit on this campus, peeping through the professors’ window and instantly feeling reaffirmed in how this space includes them.”

“I care that the technology being developed better represents society and is better for all of society,” said Sorelle Friedler, who has been teaching computer science at Haverford since 2014. “To do that, we need computer scientists who come from a variety of backgrounds. And my hope is that we are increasingly managing to train such a student body here.”