Kim receives inaugural dissertation award

Yoon Kim has received the inaugural Computer Science Outstanding Ph.D. Dissertation Award from the Harvard John A. Paulson School of Engineering and Applied Sciences.

Kim’s thesis, titled “Deep Latent Variable Models of Natural Language,” explores the combined use of deep learning and latent variable modeling to provide a better understanding of natural language processing, which is the study of computer interpretations of human language.

Latent variable models separate a learning problem into independent parts, while deep learning can learn complex, unusual functions for those parts. By combining those techniques, Kim’s approach can “understand” language significantly better than other methods.

“Yoon's graduate work has managed the rare accomplishment of producing natural language processing and machine learning research that has both become widely used within the research and industrial domain, while also maintaining a rigorous scientific perspective. During his time as a grad student, he had results that have been widely used for text classification, compression of large natural language processing models to devices like phones, and he even wrote open-source software that was deployed by major translation companies,” said former Assistant Professor of Computer Science Alexander Rush, who was Kim’s advisor at SEAS and is now an Associate Professor at Cornell Tech. “His thesis explores a challenging problem that he has been working on for several years, how to learn the latent properties underlying language structure for syntax and topical analysis, using a methodology known as deep generative models. It is a really beautiful work, the kind of thesis that will act as a reference for this topic for years to come.”

Prior to beginning his graduate work at SEAS, Kim earned bachelor’s degrees in mathematics and economics at Cornell University, a master’s degree in statistics at Columbia University, and a master’s degree in data science from New York University.

He will join the MIT Department of Electrical Engineering and Computer Science as an Assistant Professor.

“Yoon's dissertation describes a program of research that he has been carrying out over the last several years to recover the latent structure in language, and provides a large range of application of his ideas. His methods show a conceptual simplicity that belies their technical richness and mathematical sophistication,” said Stuart Shieber, James O. Welch, Jr. and Virginia B. Welch Professor of Computer Science. “He is already viewed as a leader in the field of deep learning applied to natural language having generated ideas that have become standard practice in the area. Yoon richly deserves the CS dissertation award.”

Earlier this year, Harvard computer science decided to formally recognize the most outstanding Ph.D. dissertations of the year, explained Eddie Kohler, Microsoft Professor of Computer Science and Area Chair of Computer Science. A committee of faculty chose the winner from a group of students who were nominated based on input from all graduating students’ Ph.D. committees.

The award evaluates each dissertation as a cohesive body of work, and looks at research quality, potential impact, and quality of presentation.

“We're proud of all our students, and we're proud that Yoon did this groundbreaking work at Harvard,” Kohler said.