Johan Ugander pursues better machine learning algorithms with NSF CAREER award

Machine learning (ML) algorithms increasingly mediate the decisions people make in their everyday lives, from what clothing to buy to what movie to watch to what healthcare plan to adopt.

As these ML technologies become increasingly at risk of being used against the public interest, there is a pressing need to develop tools to understand and mitigate the ways in which ML systems can exploit predictable "irrationalities" of human behavior. MS&E Professor Johan Ugander recently received a National Science Foundation (NSF) CAREER award to develop these tools in the public interest. His work aims to "get out ahead" of how machine learning systems can exploit human behavior, according to the proposal, and to develop understandings as well as defenses. He will develop theoretical foundations and applications for new ML algorithms that learn and predict human decisions descriptively from data as they are, rather than as normative theories prescribe them to be.

These new algorithms build on recent interpretations of choices as driven predominantly by pairwise interactions that Prof. Ugander has led, involving new tools from graph theory to model human decision making. The unique practical potential of Prof. Ugander's proposal stems from its ambitions to operationalize modern behavioral economics within a machine learning framework, making it possible to improve and adapt the design of large-scale web systems that people interact with on a daily basis.

Learn more about Stanford's nine recent CAREER award winners