The Undergraduate Research Program website, created by the Undergraduate Research Program at the University of Washington, is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.
Permissions beyond the scope of this license are available at exp.washington.edu/urp/about/rights.html
The Washington Research Foundation Fellowship
Eric Lei - Computer Science, Economics, Mathematics
I am currently working with Professor Carlos Guestrin, a leading expert in machine learning. My first exposure to machine learning was in junior year, when I took a class from Professor Guestrin about big data. I quickly realized how valuable machine learning was becoming, with tech companies collecting massive amounts of data from which important insights could be extracted. I became extremely interested by the rigorous approach of machine learning to modeling and by the emphasis on utilizing the newest technological advances to solve questions about data. At the moment Professor Guestrin and I are working on the improvement of recommendation engines--algorithms that detect user preferences and make product suggestions in services such as blog or video websites. The support from the WRF Fellowship will help us move forward with development and implementation of new ideas. After graduation, I hope to enter a PhD program in machine learning or join industry as a machine learning scientist.
Mentor: Carlos Guestrin, Computer Science & Engineering
Project Title: Community Detection in Social Networks
Abstract: A/B testing is a standard approach for evaluating the effect of online experiments; the goal is to estimate the "average treatment effect" of a new feature or condition by exposing a sample of the overall population to it. A drawback with A/B testing is that it is poorly suited for experiments involving social interference, when the treatment of individuals spills over to neighboring individuals in a community along an underlying social network. In this work, social networks are modeled as graphs, and communities are represented as clusters. We propose a novel methodology using graph clustering to analyze average treatment effects under social interference. We examine theoretical properties of this methodology as well as its empirical performance.