Given m muffins and s students, how can the muffins be divided evenly among the students, such that we maximize the smallest piece of muffin? In pursuit of solving this challenge, I worked with Prof. William Gasarch and a team of students creating a battery of techniques, delving into combinatorics, algorithms, mixed integer programming, and topology. I developed a method that used combinatorial approaches to derive upper bounds on the smallest piece of muffin, for certain cases of m and s, and created an algorithm to automatically generate these bounds.

Aspects of our work were published in the 9th International Conference on FUN with Algorithms (FUN 2018), a premier conference in the use of algorithms for results that are amusing, but nonetheless “scientifically profound”. I am also a co-author of the book, Mathematical Muffin Morsels, published by World Scientific.

Helping individuals maintain exercise behaviors has long been a subject of study within health informatics. After seeing extensive research on mobile apps for this purpose, we decided to investigate the potential of smart speakers, such as the Amazon Echo. Together with my advisor, Prof. Eun Kyoung Choe, and our collaborator from Microsoft Research, Dr. Bongshin Lee, we developed a study comparing the effectiveness of smart speaker skills versus mobile apps in encouraging consistent exercise behavior. As the main investigator in this project, I was instrumental in our collective decisions on our study design, applying for IRB approval, and implementing the mobile app, smart speaker skill, and database to integrate the tracking data collected by either device.

We submitted this study proposal and successfully published it in the 2018 CHI Conference on Human Factors in Computing Systems, Late Breaking Works track, where I presented a poster on our work, as the first author.

During the summer of 2018, I participated in the Indiana University-Purdue University of Indianapolis (IUPUI) Data Science NSF REU, conducting research with another undergraduate student and a graduate student under our advisor Prof. George Mohler. We further developed a technique for anomaly detection with generative adversarial networks (GANs). After finding lower-dimensional embeddings of image data with bidirectional GANs, we used the infinite-gaussian mixture model clustering algorithm to find an “anomalous” class. We applied this technique to human mobility data, successfully distinguishing different drivers based on their GPS data.

We have published and presented this work in the 5th National Symposium for NSF REU Research in Data Science, Systems, and Security as part of the 2018 IEEE Big Data Conference.