Project 511 at Center for Data Arts, Fall 2018 – Spring 2019

Using OpenCV to detect and segment poses on the Pompidou Collection of portraits and then cluster them using T-SNE in order to find unique patterns on body postures across the photos. The goal is to generate image sequences following these patterns in aesthetic ways and display as a media art installation. I worked on the pose detection and segmentation process using GNN, as well as classification for abnormal pose detection. Later I also Cooperated in the development of animation in Openframeworks for final installation

Statistics Honors Thesis, Fall 2017 – Spring 2018

I conducted a detailed EDA on the data by NYC Taxi & Limousine Commission (TLC). First worked on the optimal transport plan using the Sinkhorn-Knopp algorithm. Then I focused on the traffic flow and density analysis with density based cluster algorithms and Poisson process to look for popular destination areas in NYC during different time of the day.

Big Data Summer Institute, University of Michigan, School of Public Health, Summer 2017

I worked in the imaging team at the Big Data Summer Institute. We worked on the detection of melanomas by classifying the skin lesion pictures. My group applied machine learning algorithms to the problem. We presented our results at the final symposium and presentation.

Qualitative Data Reuse, University of Michigan School of Information, Fall 2016 – Winter 2017

Besides helping with literature review, survey design, cognitive walk through, interview record analysis and survey data analysis, I did my independent research on the patterns of the use of complimentary data in data reuse. I presented my results at the UROP symposium.

I also contributed as the second author to Trust in Qualitative Data Repositories, ASIS&T AM 2017, providing analysis on the data and composing the corresponding method part.