About

I'm a machine learning PhD student in the group of Prof. Joachim M. Buhmann at ETH Zurich, Switzerland.

Email: pletscher@inf.ethz.ch
Phone: +41 44 632 8608
Office: CAB F10.4

Research

My main research interest is centered around structured prediction. I'm mostly focusing on training such models and devising fast algorithms to do so. Solving the problem is computationally intractable in general. Many subproblems are closely related to probabilistic inference in graphical models, such as computing the partition sum.

Publications

I also try to keep my Google scholar profile up to date.

Peer Reviewed Conferences and Journals

Patrick Pletscher & Pushmeet Kohli
Learning Low-order Models for Enforcing High-order Statistics
AISTATS, 2012, to appear.

Patrick Pletscher & Cheng Soon Ong
Part & Clamp: Efficient Structured Output Learning
AISTATS, 2012, to appear.

Patrick Pletscher, Sebastian Nowozin, Pushmeet Kohli & Carsten Rother
Putting MAP back on the Map
33rd Annual Symposium of the German Association for Pattern Recognition (DAGM), 2011.
[PDF] [Publisher] [supplement] [poster] [BibTeX]

@inproceedings{Pletscher2011,
  author    = {Pletscher, Patrick and Nowozin, Sebastian and Kohli, Pushmeet and Rother, Carsten},
  title     = {Putting {MAP} back on the Map},
  booktitle = {33rd Annual Symposium of the German Association for Pattern Recognition ({DAGM})},
  year      = {2011},
  editor    = {R. Mester and M. Felsberg},
  publisher = {Springer},
  series    = {Lecture Notes in Computer Science},
  volume    = {6835},
  pages     = {111--121},
}

Patrick Pletscher, Cheng Soon Ong & Joachim M. Buhmann
Entropy and Margin Maximization for Structured Output Learning
Proceedings of the 20th European Conference on Machine Learning and Knowledge Discovery in Databases (ECML PKDD), 2010.
[PDF] [Publisher] [supplement] [talk] [poster] [BibTeX]

@inproceedings{Pletscher2010,
  author    = {Pletscher, Patrick and Ong, Cheng Soon and Buhmann, Joachim M.},
  title     = {Entropy and Margin Maximization for Structured Output Learning},
  booktitle = {Proceedings of the 20th European Conference on Machine Learning ({ECML})},
  year      = {2010},
  editor    = {Jos{\'e} L. Balc{\'a}zar and
               Francesco Bonchi and
               Aristides Gionis and
               Mich{\`e}le Sebag},
  series    = {Lecture Notes in Computer Science},
  volume    = {6321},
  pages     = {83-98},
}

Patrick Pletscher, Cheng Soon Ong & Joachim M. Buhmann
Spanning Tree Approximations for Conditional Random Fields
Proceedings of the Twelfth International Conference on Artificial Intelligence and Statistics (AISTATS), 2009.
[PDF] [Publisher] [BibTeX]

@inproceedings{Pletscher2009,
  author    = {Pletscher, Patrick and Ong, Cheng Soon and Buhmann, Joachim M.},
  title     = {Spanning Tree Approximations for Conditional Random Fields},
  booktitle = {Proceedings of the Twelfth International Conference on Artificial
               Intelligence and Statistics ({AISTATS}) 2009},
  year      = {2009},
  pages     = {408--415},
  editor    = {D. van Dyk and M. Welling},
  publisher = {JMLR: W&CP 5},
  address   = {Clearwater Beach, Florida}
}

Matthew Brand & Patrick Pletscher
A conditional random field for automatic photo editing
Proceedings of the IEEE Conferencer on Computer Vision and Pattern Recognition (CVPR), 2008.
[PDF] [Publisher] [BibTeX]

@inproceedings{Brand2008,
  author    = {Matthew Brand and Patrick Pletscher},
  title     = {A conditional random field for automatic photo editing},
  booktitle = {Proceedings of the {IEEE} Conference on Computer
               Vision and Pattern Recognition ({CVPR})},
  year      = {2008},
}

Isabelle Guyon, Jiwen Li, Theodor Mader, Patrick Pletscher, Georg Schneider & Markus Uhr
Competitive baseline methods set new standards for the NIPS 2003 feature selection benchmark
Pattern Recognition Letters, Volume 28, Issue 12, September 2007, Pages 1438-1444.
[PDF] [Publisher] [BibTeX]

@article{Guyon2007,
  author    = {Isabelle Guyon and Jiwen Li and Theodor Mader and
               Patrick A. Pletscher and Georg Schneider and Markus Uhr},
  title     = {Competitive baseline methods set new standards for the
               {NIPS} 2003 feature selection benchmark},
  journal   = {Pattern Recognition Letters},
  volume    = {28},
  number    = {12},
  year      = {2007},
  issn      = {0167-8655},
  pages     = {1438--1444},
  doi       = {http://dx.doi.org/10.1016/j.patrec.2007.02.014},
  publisher = {Elsevier Science Inc.},
  address   = {New York, NY, USA},
}

Workshops

Patrick Pletscher & Sharon Wulff
A Combined LP and QP Relaxation for MAP
NIPS Workshop on Discrete Optimization in Machine Learning (DISCML), 2011.
[PDF] [spotlight] [poster] [BibTeX]

@inproceedings{Pletscher2011a,
  author    = {Pletscher, Patrick and Wulff, Sharon},
  title     = {A Combined LP and QP Relaxation for MAP},
  booktitle = {NIPS Workshop on Discrete Optimization in Machine Learning (DISCML)},
  year      = {2011}
}

Reports

Patrick Pletscher
Model order selection: Criteria, inference strategies and an application to biclustering
Master's thesis, ETH Zurich, September 2007.
Supervised by Peter Orbanz and Prof. Buhmann.
[PDF] [BibTeX]

@mastersthesis{Pletscher2007,
  author  = {Patrick Pletscher},
  title   = {{Model order selection: Criteria, inference strategies and an application to biclustering}},
  school  = {ETH Zurich},
  address = {Switzerland},
  year    = {2007},
  month   = {September},
}

Peptide Assignment Validation
Semester thesis in machine learning, 2006.
Supervised by Bernd Fischer and Prof. Buhmann.
[PDF] [talk] [BibTeX]

@misc{Pletscher2006,
  author  = {Patrick Pletscher},
  title   = {{Peptide Assignment Validation: Telling what's wrong without
              actually knowing what's right}},
  school  = {ETH Zurich},
  address = {Switzerland},
  year    = {2006},
  month   = {April},
  note    = {semester project report},
}

Adaptive Security of Compositions
Semester thesis in cryptography, 2005.
Supervised by Krzysztof Pietrzak and Prof. Maurer.
[PDF] [talk] [BibTeX]

@misc{Pletscher2005,
  author  = {Patrick Pletscher},
  title   = {{Adaptive security of compositions}},
  school  = {ETH Zurich},
  address = {Switzerland},
  year    = {2005},
  month   = {July},
  note    = {semester project report},
}

Software

Graph cut MEX wrapper

Matlab MEX wrapper for Vladimir Kolmogorov's graph cut code.
graphcut-1.0.tar.gz (Release date: 2010/03/17).

Nonparametric Bayesian Biclustering

Nonparametric Baysian Biclustering with a Double Mixture Model. This work was carried out in parts during my master's thesis.
npbb-1.0.tar.gz (Release date: 2008/12/26).

Teaching

I have assisted several courses at ETH Zurich as a teaching assistant.

Probabilistic Graphical Models for Image Analysis WS11
Computational Intelligence Lab SS11
Probabilistic Graphical Models for Image Analysis WS10
Computational Intelligence Lab SS10
Probabilistic Graphical Models for Image Analysis WS09
Informatik II (D-MAVT) SS09
Image Analysis with Statistical Models WS08
Visual Computing SS08
Computational Science SS07
Computational Science SS06
Informatik I (D-BAUG) WS05/06

If you are an undergraduate student at ETH Zurich and looking for a research topic for e.g. a master's thesis, then feel free to contact me. Possible topics include approximate inference and learning in graphical models and its applications to e.g. computer vision.