|Class Time:||11:15am to 12:30pm|
|Pre-Requisites:||One of CSCI-B551, CSCI-B555, INFO-529, INFO-I611.|
|Syllabus:||No Syllabus Avaliable|
|Description:||The importance of relational (structured) data is evident from its increasing presence: WWW, social networks, relational databases, bibliographic networks, organizational networks, biological pathways, and many more. The rich information in relational data gives rise to a wealth of potential patterns that may characterize a network. The ability to describe and detect relational patterns provides powerful support for many applications, including social network analysis, viral marketing, information extraction, drug discovery, computer vision, robotics and many more. In this course, we will explore statistical-relational learning (SRL) methods that extend machine learning techniques so that they apply to relational domains made up of objects that interrelate. Statistical-relational systems employ probability to reason about uncertainty in network structures. They utilize the expressive power of formal logic to represent the full complexity of heterogeneous networks with multiple types of links, nodes, and attributes. In addition to learning about the different formalisms, we will also cover inference algorithms for such models. At the end of the course, the students will build a SRL system that can be used to learn from structured data and perform inference. The first half of the course will be lectures while the second half will involve presentations from students on different topics.
Any student who is looking at multi-relational data and is interested in machine learning will benefit from the learning and inference techniques that are being presented in this course.