FEAST 2015

Update 13/7/2015:

The workshop took place on the 10th July 2015 at ICML and is now over. Thanks to all the speakers and participants, and congratulations to poster prize winners Francesco Orsini and Emilie Kaufmann!

You can still check out the program (with slides), the abstracts and some pictures.

Keep an eye on this website and ml-news for future editions of this workshop!

---Chloé-Agathe Azencott, Veronika Cheplygina, and Aasa Feragen


The second workshop on Features and Structures (FEAST), colocated with ICML, aims to provide an open, low ­threshold forum for discussing data consisting of an underlying structure associated with descriptive attributes, such as in natural language processing, medical imaging, computer vision, bioinformatics, chemoinformatics, social network analysis and others. The goal of FEAST is to bring together researchers from diverse communities, with a shared interest in understanding structured data.


Room Turin at the Lille Grand Palais conference center.

Invited speakers

Robert P.W. Duin, Delft University of Technology Structural pattern recognition in the dissimilarity space

Florence d’Alché-Buc
LTCI CNRS 5141, Télécom Paristech, Université Paris-Saclay — Learning operator-valued kernels in multiple output regression

Pierre Vandergheynst, EPFL - High-dimensional signal processing on graphs

Matthew Blaschko, INRIA & École Centrale Paris — Non-modular losses in structured prediction with applications to computer vision

Vitaly Kuznetsov, Courant Institute of Mathematical Sciences - Ensemble methods for structured prediction

Motivation and topics

The goal of the workshop is to spark discussion and interaction. Submission requires a 1-page PDF abstract, and we particularly encourage:

-Previously published work

-Preliminary results

-Open-ended questions and problem formulations

We welcome contributions including but not limited to:

-structured data

-data representation

-multiple instance learning

-graph kernels

-topological descriptors

-structured output

-graph analysis

-structured sparsity

Important dates

1st May 2015 - Deadline for abstracts

6th May 2015 - Extended deadline for abstracts

10th May 2015 - Acceptance notification

15th May 2015 - ICML early registration deadline

10th July 2015 - Workshop in Lille, France

Submission requires a 1-page double-column IEEE style abstract, which may cover previously published work, ongoing work or an open problem. Selected abstracts will be published on the workshop website.


Chloé-Agathe Azencott, Centre for Computational Biology of Mines ParisTech, Institut Curie and INSERM in Paris, France

Veronika Cheplygina, Biomedical Image Group Rotterdam, Erasmus Medical Center, Rotterdam, The Netherlands.

Aasa Feragen, Department of Computer Science, University of Copenhagen, Denmark


Textkernel (http://www.textkernel.com): machine learning for matching people and jobs
Eagle Vision (http://eaglevision.nl): developing innovative vision inspection systems for several industries