Postdocs, knowledge reuse, Machine Learning, Data Mining

Previous post

Next post

Several postdoctoral positions are available to perform high-quality research on the REFRAME project (Rethinking the Essence, Flexibility and Reusability of Advanced Model Exploitation) granted under the CHIST-ERA 2011 call to a consortium consisting of the University of Bristol (project coordinator), Polytechnic University of Valencia and the University of Strasbourg.

The overall aim of the project is to develop an innovative and principled approach to knowledge reuse which will allow a range of known machine learning and data mining techniques to anticipate and deal with common contextual changes. The approach is based around the new notion of model reframing, which can be applied to inputs (features), outputs (predictions) or parts of models (patterns), in this way generalising, integrating and broadening the more traditional and diverse notions of model adjustment in machine learning and data mining. The ultimate goal of the project is to provide a much better understanding of the issues involved in the generation and deployment of a model for different contexts, as well as the development of tools which ease the extraction, reuse, exchange and adaptation of knowledge for a wide spectrum of operating contexts.

To find out more about the project, visit www.reframe-d2k.org.

PostDoc, adapting models to changes in relational data and geographical applications, Strasbourg

The university of Strasbourg is in charge of adaptations related to relational data and to changes in the background knowledge, on the one hand, and of applications to geographical problems, on the other hand. We are looking for a post-doctoral fellow with a relevant background in data mining or machine learning, with experience in cost-sensitive learning/imbalanced data, in relational data mining/inductive logic programming/statistical relational learning, or in geomatics. French speaking is not mandatory for work.

The position is available immediately, and can be extended yearly, up to three years. Applications should be submitted as soon as possible, and will be considered until the position is filled.

Enquiries and applications, including CV and references, must be sent to Nicolas Lachiche, nicolas.lachiche@unistra.fr.

PostDoc, generalisation of ROC analysis and applications in the smart electricity meter domain, Bristol

Part of the work at the University of Bristol aims to extend ROC (Receiver Operating Characteristic) analysis beyond the restricted view of operating condition as a class distribution or a cost proportion. The notions of dominance, operating point, threshold choice, probability calibration, and aggregated performance metrics such as AUC can be generalised to a much wider range of operating contexts. The generalisation to non-binary and real-valued target variables, input variables of various types, and feature ensembles and hierarchies provide concrete starting points for this work.

We are looking for an experienced post-doctoral research assistant with a relevant background in data mining or machine learning, with experience in cost-sensitive learning and ROC analysis. The successful candidate will also be in charge of maintaining a data set of readings from smart electricity meters that provides one of the application domains of the project, and to contribute to the overall management of the project.

The position is available immediately and is for 2.5 years.

For informal enquiries contact Peter Flach, Peter.Flach@bristol.ac.uk;

Further details of the position and a link to the online application system are at www.bris.ac.uk/jobs/find/details.html?nPostingTargetID=1030.

The closing date for applications is 19 December 2012.

PostDoc, reframing in the context of incompleteness, Bristol
Other work at the University of Bristol concerns adapting existing techniques for nonmonotonic and abductive inference in order to deal with incompleteness in relational background knowledge. Initially this will involve abductive techniques for dealing with missing data. This will then be developed into a generic abductive-inductive technique for knowledge reframing in the context of incompleteness.

We are looking for an experienced post-doctoral research assistant with a relevant background in data mining or machine learning, with experience in relational data mining/inductive logic programming.

This position will be available from late 2013 and is for 1.5 years. Informal enquiries and applications, including CV and references, can be sent to Peter Flach, Peter.Flach@bristol.ac.uk; details of the formal application procedure will be posted in due course.

Previous post

Next post