Contributing to New Knowledge
On January 18, Northwestern CS hosted the inaugural Undergraduate Research Showcase
Northwestern Engineering’s Samir Khuller welcomed the presenters and guests of the inaugural Undergraduate Research Showcase with an anecdote of a conversation he had with a young relative about what professors do besides teach. Khuller explained to his nephew that he conducts research — inventing new knowledge in a continually evolving process.
“That blew his mind,” said Khuller, Peter and Adrienne Barris Chair of Computer Science at the McCormick School of Engineering. “He thought every piece of knowledge is contained in textbooks.”
On January 18, Northwestern CS invited six undergraduate students to present their exemplary contributions to new knowledge. Sara Sood, Chookaszian Family Teaching Professor of Computer Science and associate chair for undergraduate education at Northwestern CS, organized and moderated the virtual event, which drew 43 attendees, including undergraduate students, faculty, staff, and Dean Julio M. Ottino.
“Our students not only shared their research contributions, they also demonstrated the breadth of the field of computer science,” Sood said. “I hope they inspire other undergraduates to pursue research opportunities.”
Liam O'Carroll
Liam O'Carroll, a third-year student in Northwestern CS, presented “The Burer-Monteiro Method Can Fail Above the Barvinok-Pataki Bound,” a joint project with Vaidehi Srinivas, a first-year PhD student in computer science, and Aravindan Vijayaraghavan, associate professor of computer science and (by courtesy) industrial engineering and management sciences at Northwestern Engineering.
O’Carroll met Vijayaraghavan when he took the COMP_SCI 212: Mathematical Foundations of Computer Science course during his first year. O’Carroll approached Vijayaraghavan at the conclusion of the course and expressed interest in continuing to learn the material. O’Carroll subsequently enrolled in several advanced theory graduate courses and was a top performer.
The focus of O’Carroll’s research centers around the memory-optimality trade-off in semidefinite programs (SDP), a powerful and important class of mathematical optimization models widely applicable in control systems, signal processing, machine learning, computational biology, and statistics. He discussed a well-known application of SDP called the Max-Cut Problem and explained how the problem of finding the maximum cut in a graph with potentially millions of edges of vertices can be encoded as a Burer-Montero optimization problem over an array of numbers.
“Memory usage quickly becomes a bottleneck for solving large-scale SDPs,” O’Carroll said. “If you can solve the Burer-Montero problem then you can instantly obtain a solution to the original SDP using less memory.”
“Liam operates at the level of a PhD student,” Vijayaraghavan said. “I learned much more from him than he has from me on this project.”
O’Carroll encouraged fellow undergraduates to attend seminars and events in the department, such as the Algorithms Coffee gatherings held weekly at 3 p.m. Wednesday in the Theory Collaboration Area on the third floor of Mudd Hall.
“Don’t be afraid to reach out to professors,” O’Carroll said. “Don’t be intimidated – reach out.”
Aldo Aguilar
Aldo Aguilar, a fourth-year undergraduate student, initially joined Northwestern as a saxophone performance and mathematics major. After switching to the computer science program in Northwestern Engineering, Aguilar joined the Interactive Audio Lab (IAL). Bryan Pardo, professor of computer science in Northwestern Engineering and of radio/television/film in the School of Communication, described the move as “stealth mode.”
Aguilar first approached and connected with Hugo Flores García, a PhD student in computer science in IAL, who suggested to Pardo that they should invite Aguilar to a meeting. At that time, Aguilar had worked as an undergraduate research assistant and had taken 300-level CS courses but did not have much experience with the audio field.
“Before I knew it: he’s in my lab.” said Pardo, a co-director of the Center for Human-Computer Interaction + Design. “He’s been a joy to work with. Aldo is a great contributor and very community-minded in his approach to doing work.”
Aguilar, Flores García, Pardo, and Ethan Manilow, PhD candidate in computer science, recently won the Best Paper Award at the International Society for Music Information Retrieval (ISMIR) Conference for their development of a new method of hierarchical prototypical networks to identify a more diverse range of musical instruments in a machine learning audio recognition mechanism.
Aguilar presented the second project he published with the IAL team — “Deep Learning Tools for Audacity” — that connects deep learning audio domain researchers directly with end users of sound design tools, including speech detection, instrument labeling, and audio upscaling. The group created a framework for the digital audio workstation Audacity, a free, accessible, open-source audio editor, which allows researchers to easily share their models with users, both enriching the audio processing toolkit and streamlining otherwise time-intensive tasks. The team submitted the paper in November 2021, just six months after launching the project.
Marc Jiang and Sydney Simmons
Marc Jiang and Sydney Simmons are members of the Technological Innovations for Inclusive Learning and Teaching (tiilt) Lab, which aims to improve learning opportunities for students from underserved communities. tiilt is led by Marcelo Worsley, assistant professor of computer science at the McCormick School of Engineering and of learning sciences at Northwestern’s School of Education and Social Policy.
Jiang and Simmons are part of tiilt’s Making a Just, Inclusive Community (MAJIC) team, focused on working with teachers and community organizations to study learning spaces and apply computation methods to explore and develop new learning methods.
Jiang presented the Famjam project, which facilitates meaningful learning and “making” experiences for families. Family members are introduced to tools in the lab, including the laser cutter and 3D printer, and are encouraged to incorporate these technologies into the creation of a game or activity that reflects their collective identity and shared narrative. One family, for instance, designed a custom-version of the Parker Brothers game Sorry!. Famjam applies computational network analysis to track the spatial orientation of family members during the making process, which helps the tiilt team study learning patterns.
Simmons discussed the MAJIC team’s professional development work with teachers in Evanston/Skokie School District 65 inspired by the concept of inclusive making, or the matching of learning experiences to the needs and participation styles of individual students. MAJIC supports District 65 throughout the school year by assisting educators in securing funding for fabrication tools, hosting extracurricular events to help students learn academic concepts in engaging ways, and providing care packages to the teachers. The team analyzes their impact through traditional qualitative research methods as well as text mining of social media posts related to the maker activities.
“It's been really exciting for me to watch the two of them grow through the research process,” Worsley said. “Marc and Sydney learned how to engage with the community and also to think about ways that the computer science work they’ve been doing can extend and grow into community spaces.”
Jiang and Simmons also provide training to college interns at World Relief Chicago, an organization that hosts summer camps for refugee children, on how to facilitate social-emotional learning.
“MAJIC has been fundamental to my time at Northwestern in terms of figuring out what I want to do in the future,” Jiang said. “It’s encouraging to see how computer science can be applied in such broad ways to enhance the way people do things that matter to them.”
Peter Zhong
Peter Zhong is a third-year computer science student in the Programming Languages Lab. He received an honorable mention in the Computer Research Association (CRA) 2022 Outstanding Undergraduate Researcher Awards for his work on the theoretical foundations of undefined behavior, a facet of modern compiler technology.
Zhong collaborated on the project — “Reaching the Unreachable - Exploring How an Optimizing Compiler Propagate and Eliminate Unreachable Code” — with Shuhung You, a PhD student in computer science; Robby Findler, associate chair and professor of computer science; Christos Dimoulas, assistant professor of computer science; and Simone Campanoni, assistant professor of computer science and (by courtesy) electrical and computer engineering.
Compiler optimizations determine always-true facts about programs written in modern languages, including region of codes that should never be executed, termed unreachable. Zhong presented a set of transformation rules that define how a compiler propagates unreachability facts to mark code segments and eliminate dead code. He then outlined a correctness proof and theorem addressing confidence in the performance of modern compilers.
Ottino asked the group how common it is for undergraduate researchers to develop theorems and inquired about the significance of Zhong’s theorem.
Findler acknowledged that many undergraduates do produce theorems in programming languages and other theoretical topics.
“The proof techniques that Peter had to master to be able to prove this result are non-trivial. I was an assistant professor when I first really got my head around the techniques that Peter picked up in a month or two,” Findler said. “The difficulty of this theorem is high and the significance is yet to be understood. The general area of dealing with undefined behavior and capturing the correctness of compiler translations like this is clearly important.”
“We’re studying real-world applications to see how valid and applicable this result is to systems that millions of people actually use every day,” Zhong said.
Shirley Zhang
Shirley Zhang completed her undergraduate degree at Princeton University in 2020 and is currently applying to graduate school. She contacted Khuller to discuss research opportunities and, after a few coffee shop meetings, joined a project with Sami Davies, a National Science Foundation Computing Innovation Fellow and postdoctoral researcher in theoretical computer science — “On minimizing average flow time for batch scheduling.”
Using the example of a shuttle company picking up passengers from the airport, Zhang explained the problem of minimizing the average flow time of jobs subject to an active time constraint. She then presented a dynamic program which optimally schedules unit jobs to minimize flow time given a maximum allowed number of active time slots. Zhang also described future directions for the project, including non-agreeable deadlines, weighted jobs, and online settings.
Additional presentations
Khuller expressed his hope that the Undergraduate Research Showcase event will inspire more undergraduate students to consider starting research projects with Northwestern CS faculty as well as pursuing graduate school.
Additional undergraduates will present their research in part two of the showcase. The date and time will be determined.