Challenges in Structured Models
Date/Time: December 3rd, 3:40 pm
Location: 1312 Hoover
Two fundamental ideas in data science are modeling (representing some phenomenon) and loss functions (representing an agent's priorities). While structured models have shown great potential in problems ranging from natural language processing to computational biology to computer vision, computational difficulties impede their use. After an introduction to data science and structured models, this talk will discuss how modeling, computation, and loss functions interact when building and querying structured models.
Justin Domke received a PhD in computer science from the University of Maryland in 2009. From 2009 to 2012, he was an Assistant Professor at Rochester Institute of Technology. Since 2012, he is a member of the Machine Learning group at NICTA and an adjunct at the Australian National University.