CVPR2014: Learning and Inference in Discrete Graphical Models

Organizers

Nikos Komodakis (Ecole des Ponts ParisTech, Universite Paris-Est)
Nikos Paragios (Ecole Centrale de Paris, INRIA-SACLAY)

Stephen Gould (Australian National University)
Matthew Blaschko (Ecole Centrale de Paris, INRIA-SACLAY)
Pushmeet Kohli (Microsoft Research Cambridge)
Karteek Alahari (INRIA Grenoble)
Dhruv Batra (Virginia Tech)

Opening & Introduction

Part I:	Inference
[S. Gould, 45 min]	Exact inference in graphical models	[slides]
	Graph-cut based methods
	Relaxations and dual-decomposition
[P. Kohli, 45 min]	Strategies for higher-order models	[slides]
[D. Batra, 15 min]	M-Best MAP, Diverse M-Best	[slides]

Part II:	Learning	[slides] (by N. Komodakis)
[M. Blaschko, 45 min]	Introduction to learning of graphical models
	Maximum-likelihood learning, max-margin learning
	Max-margin training via subgradient methods
[K. Alahari, 45 min]	Constraint generation approaches for structured-output learning
	Efficient training of graphical models via dual-decomposition

Location: Greater Columbus Convention Center, C115
Date: June 28