Yimin Tan
Master student
Department of Computer Sciences Email: aca3tu@r.postjobfree.com
University of Wisconsin-Madison Telephone: +1-608-***-****
Objective
Software Engineering (specialized in Machine learning, Data mining)
Education
University of Wisconsin-Madison, Computer sciences Department GPA:3.8
•
Master student,Research assistant Advisor: Prof Xiaojin(Jerry) Zhu Sept. 2010 – Dec. 2013
Tsinghua University, Electronic engineering Department GPA:90.2
•
B.E. undergraduate student(graduate with honor) Sept. 2006 – July 2010
Selected Machine learning related courses
CS: Machine learning, Advanced Machine learning, Data model, Nonlinear optimization
Stats: Mathematical statistics i,ii ECE: Statistical signal processing
Internship
Prediction of user scrolling preference in News-feed May 2012 – Aug. 2012
•
Software Engineering, Facebook Inc. (News-feed ranking team)
– Collected users’ scrolling behavior when going through facebook page (Javascript)
– Data logging, fetching, joining on server-side (PHP,SQL)
– Designed relevant features, trained prediction models,and analyzed model performance offline (C++)
– Run online A/B test, and improved News-Feed Ranking system based on scrolling preference model
Research experience
• Spatial-Temporal signal recovery from Twitter data Nov. 2011 – Sept. 2012
– Proposed probabilistic model to reconstruct spatial-temporal maps of events from social media (Twitter data)
– Formulated signal recovery as Poisson intensity estimation with regularization
• Safe semi-supervised bagging Sept. 2012 – June 2013
– Proposed density-ratio bagging, a safe semi-supervised extension of bagging
– Designed synthetic and real experiments on variety of semi-supervised regression and classification algorithms
• Anti/Pro semi-supervised learning (SSL) method Jan. 2011 – Oct. 2011
– Systematically analyzed situation where SSL succeeds or fails
– Constructed anti/pro-SSL dataset pairs for existing SSL algorithms, which satisfy/violate underlying SSL
assumptions,and proved their success or failure with SSL
• Bayesian graphical model for Topic modeling Oct. 2009 – May 2010
– Proposed Topic-weak-correlated Latent Dirichlet Allocation (TWC-LDA) for topic modeling, which constrains
different topics to be weak-correlated
– TWC-LDA successfully discovers weak-correlated topics with distinctive semantic meanings
Publication
Yimin Tan, Xiaojin Zhu. Dragging:density-ratio bagging UW-Madison CS Technical report TR1795, 2013
Yimin Tan, Zhijian Ou. Topic-weak-correlated Latent Dirichlet Allocation
In Proc. International Symposium on Chinese Spoken Language Processing (ISCSLP), Taiwan, 2010.
Technical Skills
Programming languages C/C++, Python, Java, PHP, Javascript, Matlab