Institute of Information Science, Academia Sinica



Press Ctrl+P to print from browser


Learning to Predict Complex Structures with Indirect Supervision

  • LecturerMr. Ming-Wei Chang (Ph.D. Candidate, Computer Science Department, University of Illinois at Urbana-Champaign)
    Host: Dr. Tyng-Luh Liu
  • Time2011-01-20 (Thu.) 10:30 – 12:00
  • LocationAuditorium 106 at new IIS Building



Machine learning techniques have made a lot of progress recently. 
Nevertheless, the current machine learning techniques only support 
limited learning protocols. Since there are only restricted ways to 
teach machines, we cannot transfer human knowledge to machines 
On the other hand, structured tasks, which involve many 
interdependent decisions for a given example, are expensive to label. 
Given that almost all natural language processing tasks are 
structured tasks, it is important to have learning frameworks that 
uses resources in addition to labeled examples.
This talk addresses the problem of reducing the labeling cost for 
structural tasks. We develop advanced machine learning algorithms 
that take advantage of indirect supervision together with labeled 
data. Indirect supervision can come in the form of constraints or 
weaker supervision signals. Our proposed learning frameworks can 
handle both structured output problems and problems with latent 
structures. We demonstrate the effectiveness of our indirect 
supervision frameworks for various natural language processing 




Ming-Wei Chang is a Ph.D. candidate in University of Illinois at 
Urbana-Champaign. His research interest is in machine learning and 
its applications to natural language processing.
Ming-Wei has published several papers on leading machine learning 
and natural language processing conferences, including ICML, ACL, 
EMNLP, NAACL, KDD and AAAI. In 2009 and 2010, he co-presented 
tutorials on combining human knowledge with statistical models in 
EACL and NAACL, respectively. Before coming to the United States, 
Ming-Wei joined several projects on support vector machines with 
Chih-Jen Lin in National Taiwan University. Together with Chih-Jen 
Lin, he won the first place in two international machine learning