Perceiving Systems, Computer Vision

Preserving Modes and Messages via Diverse Particle Selection

2014

Conference Paper

ps


In applications of graphical models arising in domains such as computer vision and signal processing, we often seek the most likely configurations of high-dimensional, continuous variables. We develop a particle-based max-product algorithm which maintains a diverse set of posterior mode hypotheses, and is robust to initialization. At each iteration, the set of hypotheses at each node is augmented via stochastic proposals, and then reduced via an efficient selection algorithm. The integer program underlying our optimization-based particle selection minimizes errors in subsequent max-product message updates. This objective automatically encourages diversity in the maintained hypotheses, without requiring tuning of application-specific distances among hypotheses. By avoiding the stochastic resampling steps underlying particle sum-product algorithms, we also avoid common degeneracies where particles collapse onto a single hypothesis. Our approach significantly outperforms previous particle-based algorithms in experiments focusing on the estimation of human pose from single images.

Author(s): Jason Pacheco and Silvia Zuffi and Michael J. Black and Erik Sudderth
Book Title: Proceedings of the 31st International Conference on Machine Learning (ICML-14)
Volume: 32
Number (issue): 1
Pages: 1152-1160
Year: 2014
Month: June
Publisher: J. Machine Learning Research Workshop and Conf. and Proc.

Department(s): Perceiving Systems
Research Project(s): 2D Pose from Images
Part-based Body Models
Bibtex Type: Conference Paper (inproceedings)
Paper Type: Conference

Event Name: International Conference on Machine Learning (ICML)
Event Place: Beijing, China

Address: Beijing, China
URL: http://jmlr.org/proceedings/papers/v32/

Links: pdf
SupMat

BibTex

@inproceedings{Pacheco:ICML:2014,
  title = {Preserving Modes and Messages via Diverse Particle Selection},
  author = {Pacheco, Jason and Zuffi, Silvia and Black, Michael J. and Sudderth, Erik},
  booktitle = {Proceedings of the 31st International Conference on Machine Learning (ICML-14)},
  volume = {32},
  number = {1},
  pages = {1152-1160},
  publisher = {J. Machine Learning Research Workshop and Conf. and Proc.},
  address = {Beijing, China},
  month = jun,
  year = {2014},
  doi = {},
  url = {http://jmlr.org/proceedings/papers/v32/},
  month_numeric = {6}
}