BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Adaptive and Move Making Auxiliary Cuts for Binary Pairwise Energi
 es - Olga Veksler (University of Western Ontario)
DTSTART:20170908T110000Z
DTEND:20170908T115000Z
UID:TALK78481@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:<span>Co-author: Lena Gorelick		(University of Western Ontario
 )        <br></span><span><br>Many computer vision problems require optimi
 zation of binary non-submodular energies. In this context\, local iterativ
 e submodularization techniques based on trust region (LSA-TR) and auxiliar
 y functions (LSA-AUX) have been recently proposed.  They achieve state-of-
 the-art-results on a number of computer vision applications. We extend the
  LSA-AUX framework in two directions. First\, unlike LSA-AUX\, which selec
 ts auxiliary functions based solely on the current solution\, we propose t
 o incorporate several additional criteria. This results in tighter bounds 
 for configurations that are more likely or closer to the current solution.
  Second\, we propose move-making extensions of LSA-AUX which achieve tight
 er bounds by restricting the search space. Finally\, we evaluate our metho
 ds on several applications. We show that for each application at least one
  of our extensions significantly outperforms the original LSA-AUX. Moreove
 r\, the best extension of LSA-AUX is comparable to or better than LSA-TR o
 n four out of six applications.</span>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR