BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Sanjeeb Dash (IBM Research)
DTSTART:20210525T160000Z
DTEND:20210525T163000Z
DTSTAMP:20260419T092329Z
UID:MIP2021/7
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/MIP2021/7/">
 Integer and linear programming methods for two problems in machine learnin
 g</a>\nby Sanjeeb Dash (IBM Research) as part of Mixed Integer Programming
  Workshop 2021\n\n\nAbstract\nIn this talk we describe linear and integer 
 programming based methods for two problems in statistics and machine learn
 ing. In the first part\, we focus on learning causal structures in the pre
 sence of latent variables\, which can be modeled as the problem of optimiz
 ing over the set of acyclic directed mixed graphs (containing directed and
  bidirected edges) defined on a set of nodes. We give an integer programmi
 ng formulation of this problem\, and the first algorithm to solve this pro
 blem to optimality (via branch-and-cut). In the second part of the talk\, 
 we will describe a linear programming based approach to learning first-ord
 er logical rules for the knowledge graph link prediction problem.\nThis is
  joint work with Rui Chen\, Tian Gao\, and Joao Goncalves.\n
LOCATION:https://researchseminars.org/talk/MIP2021/7/
END:VEVENT
END:VCALENDAR