IJCAI’15 Workshop on Sensitivity Analysis and Robustness in Probabilistic Graphical Models

Buenos Aires, July 25-27, 2015 - http://ipg.idsia.ch/wijcai15/ SECOND CALL FOR PAPERS

Probabilistic graphical models are important tools in machine learning and artificial intelligence for reasoning with uncertainty. They provide means to represent large multivariate domains compactly and to perform sophisticated learning and reasoning efficiently. Examples of probabilistic graphical models are Bayesian networks, Markov Random fields, chain and factor graphs, Gaussian graphical models, to name but a few. The quantification of these models usually requires sharp (i.e., precise) assessments of the model local potentials and might be subject to robustness issues. For instance, perturbations of some parameter values may lead to different decisions from those which would be achieved by the unperturbed model, suggesting that decisions are not reliable. Reliability might also be in question because of missing data and assumptions behind the process.

The workshop invites submissions of papers on all aspects of sensitivity analysis and robustness in probabilistic graphical models. Contributions may have a theoretical focus and/or an applied focus. A non-exhaustive list of topics follows.

• Local and/or global sensitivity analysis.
• Parameter-based and/or decision-based sensitivity analysis.
• Design of robust learning, inference and/or decision making approaches.
• Robust analysis and design of robustness measurements.
• Extensions of probabilistic graphical models.
• Reliable qualitative learning and reasoning.
• Robust treatment of missing data.
• Imprecise probability and other theories related to sensitivity analysis.
• Computational complexity, exact and approximate algorithms.

Each submission will be reviewed by peers using a double-blind process (please use the third person in self citations and take all necessary care not to identify yourselves). Accepted papers will be published electronically in a volume of the JMLR Workshop and Conference Proceedings series. There will be no rebuttal phase, but contributions considered worth publishing and needing substantial revision might be subject to a second round of reviewing/evaluation. All accepted papers will be presented at the workshop. At least one of the paper's authors should register and attend the workshop to present the work.

Submissions must be formatted according to style and template files available for the Journal of Machine Learning Research (JMLR) Workshop and Conference Proceedings - two-column version. The files are available at http://ipg.idsia.ch/wijcai15/sarpgm15.tar.gz. Papers (including figures, tables, references, etc) are expected to have between 6 and 10 pages.

IMPORTANT DATES Apr 27, 2015 - Deadline for submissions of contributions May 20, 2015 - Workshop paper acceptance notification May 30, 2015 - Deadline for workshop camera-ready copy (in case of minor revision; contributions needing major revision might need additional time - this will be arranged case by case)

PC MEMBERS Alessandro Antonucci*, IDSIA, Switzerland. Alessio Benavoli, IDSIA, Switzerland. Cassio P. de Campos*, Queen's University Belfast, UK. Arthur Choi, University of California, Los Angeles, USA. Giorgio Corani*, IDSIA, Switzerland. Fabio Cozman, University of Sao Paulo, Brazil. Adnan Darwiche, University of California, Los Angeles, USA. Sebastien Destercke, Univ. de Technologie de Compiegne, France. Marek Druzdzel, University of Pittsburgh, USA. Johan Kwisthout, Radboud University Nijmegen, The Netherlands. Agnieszka Onisko, Bialystok University of Technology, Poland. Denis Maua, University of Sao Paulo, Brazil. Serafin Moral, Universidad de Granada, Spain. Silja Renooij, Universiteit Utrecht, The Netherlands. Matthias Troffaes, University of Durham, UK. (*: Workshop organizers.)

More details about the submission procedure are available online. http://ipg.idsia.ch/wijcai15/ ++++++++++++++++++++++++++++++++++++++++ (We apologize in case you receive multiple copies of this announcement, but yet we hope to reach the greatest possible number of people. Finding a trade-off is not an easy task.)

Alessandro Antonucci IDSIA Dalle Molle Institute for Artificial Intelligence Via Cantonale (Galleria 2) CH-6928, Manno-Lugano, CH

mail: alessandro(a)idsia.ch skype: alessandro.antonucci tel: +41 916108515 web: www.idsia.ch/~alessandro