My broad interests are data privacy, differential privacy, machine learning, data mining, and graph computation.
For more detail about me, please refer to my resume.
Ph.D. candidate in Computer Science, Purdue University (Aug. 2011 - present)
Dissertation title: Practical Differential Privacy for High-dimensional and Graph Data
Advisor: Prof. Ninghui Li
M.S. in Computer Science & Information Engineering, National Taiwan University (Sep. 2008 - Jul. 2010)
B.B.A in Information Management, National Taiwan University (Sep. 2004 - Jun. 2008)
Research Intern, Samsung Research America, Spring 2016
Worked with Security and Privacy team and helped support Samsung Pay analytics.
Software Engineer PhD Intern, Google, Summer 2015
Worked with Google Ads Quality team and developed a component about validating ads data.
Software Engineer PhD Intern, Google, Summer 2014
Worked with Google Cloud Dataflow team and developed a Datastore connector within Dataflow SDK.
Research Assistant, Dept. of Computer Science, Purdue University (Jan. 2014 - present)
Conduct research on data privacy.
Teaching Assistant, Dept. of Computer Science, Purdue University (Jan. 2012 - May 2014)
Teach labs and grade homework/projects on graduate (CS590) and undergraduate level courses (CS182, CS251, CS252).
Research Assistant, Dept. of Computer Science, Purdue University (Sep. 2011 - Dec. 2011)
Research Assistant, Dept. of Computer Science and Information Engineering, National Taiwan University, Taipei, Taiwan (September 2008 - July 2010)
Graph Data Publishing under Node-differential Privacy
Graph privacy has been investigated by researchers for years. Researchers developed several technique under edge-differential privacy, while applications on node-differential privacy still remains a challenging problem today, mainly because of the natural property of huge sensitivity from a graph. In this project, we tackle the graph publishing problem under node-differential privacy using bounded sensitivity. We first investigate the problem of releasing degree sequence, and then use the degree sequence to re-generate a synthetic graph. The experiments show that our bounded sensitivity technique can approximate the original distribution of degree sequence with small error.
Differential Privacy with High-dimensional Data
Satisfying differential privacy often requires large amount of noises injected in many high dimensional dataset, such as movie-rating (where a person may rate as many movies as possible) or graph data (where a person may connect every other person). However, by investigating the property of dataset, we may only inject a limited amount of noise to make published dataset private enough, rather than inject overestimated amount of noise. We call this process as sensitivity control. This project investigates the possibility for sensitivity control to tackle private data publishing problem on high-dimensional data, such as recommendation system data.
Data privacy, privacy-preserving data publishing, differential privacy, data mining and statistical data analysis, and graph computation.
Wei-Yen Day, Dong Su, Ninghui Li, and Min Lyu, "High-dimensional Data Classification with Differential Privacy", in submission.
Wei-Yen Day, Ninghui Li, and Min Lyu, "Publishing Graph Degree Distribution with Node Differential Privacy", in SIGMOD'16, San Francisco, USA, 2016. [pdf][slides]
Wei-Yen Day and Ninghui Li, "Differentially Private Publishing of High-dimensional Data Using Sensitivity Control", in the 10th ACM Symposium on Information, Computer and Communications Security (AsiaCCS), Singapore, Singapore, 2015. [pdf][slides]
Wei-Yen Day, "Visualizing Image Query Senses by Social Tags," Master Thesis, National Taiwan University, 2010.
Wei-Yen Day and Pu-Jen Cheng, "Visualizing Image Query Senses by Social Tags," in the 33rd SIGIR Conference (SIGIR 2010) query representation and understanding workshop, Geneva, Switerland. [pdf]
Wei-Yen Day, Chun-Yi Chi, Ruey-Cheng Chen, and Pu-Jen Cheng, "Sampling the Web as Training Data for Text Classification," in International Journal of Digital Library Systems (IJDLS), 2009. [pdf]
Excellent TA award on system programming course in Dept. of Computer Science and Information Engineering, National Taiwan University, Spring 2009
President Award (top 5 % academic achievement) in Dept. of Information Management, National Taiwan University, Fall 2007.