VLS Meeting, tomorrow, 19 October
Ben Herbst
kLog: A Language for Logical and Relational Learning with Kernels
I will begin the talk by motivating the importance of relational data. In particular, I will refer to scene classification, a problem in computer vision and image understanding. Next I will give a brief overview of statistical relational learning, which provides a framework for dealing with relational data. The rest of the talk is devoted to kLog, a framework for learning with relational data using kernels. I will emphasize the importance of declarative languages in machine learning, and will take some time to discuss graph kernels. After the discussion of kLog, I will show how this framework can be applied to the scene classification problem. If time permits, I will also show examples of social network analysis.
This talk will be presented as two lectures (one on Friday and the other one next week Friday).
Keywords: statistical relational learning, graph kernels, linear classifiers, logic and relational learning, scene classification, image understanding
Coffee immediately afterwards.
Ben Herbst
Applied Mathematics
University of Stellenbosch
Matieland 7602
South Africa
Phone: +27 (0)21 808-4217
Fax: +27 (0)21 808-3778
Cell Phone: +27 (0)83 566-4466