A Caltech Library Service

Inference by Learning: Speeding-up Graphical Model Optimization via a Coarse-to-Fine Cascade of Pruning Classifiers

Conejo, Bruno and Komodakis, Nikos and Leprince, Sebastien and Avouac, Jean Philippe (2014) Inference by Learning: Speeding-up Graphical Model Optimization via a Coarse-to-Fine Cascade of Pruning Classifiers. In: Advances in Neural Information Processing Systems 27 (NIPS 2014). Advances in Neural Information Processing Systems. No.27. Neural Information Processing Systems , La Jolla, CA.

[img] PDF - Published Version
See Usage Policy.


Use this Persistent URL to link to this item:


We propose a general and versatile framework that significantly speeds-up graphical model optimization while maintaining an excellent solution accuracy. The proposed approach, refereed as Inference by Learning or in short as IbyL, relies on a multi-scale pruning scheme that progressively reduces the solution space by use of a coarse-to-fine cascade of learnt classifiers. We thoroughly experiment with classic computer vision related MRF problems, where our novel framework constantly yields a significant time speed-up (with respect to the most efficient inference methods) and obtains a more accurate solution than directly optimizing the MRF. We make our code available on-line [4].

Item Type:Book Section
Related URLs:
URLURL TypeDescription
Leprince, Sebastien0000-0003-4555-8975
Avouac, Jean Philippe0000-0002-3060-8442
Additional Information:© 2014 Neural Information Processing Systems. This work was supported by USGS through the Measurements of surface ruptures produced by continental earthquakes from optical imagery and LiDAR project (USGS Award G13AP00037), the Terrestrial Hazard Observation and Reporting Center of Caltech, and the Moore foundation through the Advanced Earth Surface Observation Project (AESOP Grant 2808).
Funding AgencyGrant Number
Gordon and Betty Moore Foundation2808
Series Name:Advances in Neural Information Processing Systems
Issue or Number:27
Record Number:CaltechAUTHORS:20160401-171052390
Persistent URL:
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:65865
Deposited By: Kristin Buxton
Deposited On:04 Apr 2016 23:02
Last Modified:09 Mar 2020 13:19

Repository Staff Only: item control page