Date:  Mehryar Mohri, CIMS

Learning to Rank
  

  Ranking is a critical task in a variety of modern large-scale applications, including the design of search engines and information extraction systems. This talk presents an introduction to the learning problem of ranking. It also describes some recent algorithms, including a novel preference-based technique based on the familiar QuickSort, which benefits from remarkable theoretical guarantees.

(part of this talk is joint work with Nir Ailon, Corinna Cortes, and Ashish Rastogi).