BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Visual Inference Problems: Structured Outputs and Inverse Graphics
DTSTART:20161214T100000
DTEND:20161214T110000
DTSTAMP:20260407T144151Z
UID:c6a943869b5f51ca6fe8724239de8ce12186de8630afe881cbbf8050
CATEGORIES:Conferences - Seminars
DESCRIPTION:Abstract: Images that we perceive or record are the result of 
 interactions between light and physical objects in the world. These intera
 ctions are well understood and can accurately be modeled in modern compute
 r graphics engines. This can be understood as  simulation processes that 
 encode our understanding of the generative process. While the simulation p
 rocess is well described and understood\, the inverse of this simulator\, 
 e.g.\, to infer physical and semantic properties of scenes\, is largely un
 solved. My research aims to describe the process of visual inference in a 
 precise mathematical formulation. \n \nIn this talk I will give an overv
 iew of our recent work in this domain. I will present an informed sampling
  approach that recovers accurate posterior distributions for visual infere
 nce processes. A second work I will present is the extension of convolutio
 nal layers of deep convolutional networks. I will present a generalization
  of the bilateral filter and show how this results in a sparse and high-di
 mensional filtering operation that can be used in deep CNNs. This has inte
 resting consequences on a range of application domains. I will discuss sev
 eral applications including material estimation and also present our lates
 t results in the recovery of detailed 3D human pose from static images. \
 n\nBiography: Dr. Peter Gehler is a research group leader at the Bernstein
  Center of Computational Neuroscience (BCCN) of the University of Tübinge
 n and the Max Planck Institute for Intelligent Systems. His work lies at t
 he intersection of machine learning and computer vision. His focus is to d
 evelop scene representations that allow a detailed understanding of images
  and videos. This is a structured output prediction problem and leads to c
 hallenging inference problems for which he develops new algorithms.\nDr. G
 ehler studied computer science at the University of Bielefeld and did his 
 PhD work at the Empirical Inference group at the Max Planck Institute for 
 Biological Cybernetics. He was a postdoctoral researcher at ETH Zurich\, t
 emporary Professor at TU Darmstadt\, and Junior Research Group Leader at t
 he Max Planck Institute for Informatik in Saarbrücken. In 2014 he was the
  GCPR program chair. He is an associate editor of PAMI\, and serves as are
 a chairs for NIPS\, ICML\, ECCV\, and ICCV. 
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR