PhD position on Machine Learning for representation learning (closed)

We are looking for an excellent candidate who will undertake a PhD on the development of new machine learning methods for non-parametric representation learning with an emphasis on two-level instance description problems and recommendation problems.

In the standard propositional learning scenario instances are described by a fixed set of attributes, which take values from some domain. However in many applications the attributes have also their own descriptions. One of the classical examples is diagnostic problems in genomics and protemics in which individuals are described by the expression levels of sets of genes and proteins, which in their turn are described by their properties. The typical learning methods do not make use of the available side information. Similar settings appear in recommender problems in which items and users are also described by side information. The goal of the project is to develop representation learning methods that work in a two level setting, the base instance representation level and the second level that contains the side information on the attributes, to learn latent representations exploiting all the available information. To that end the applicant is expected to build on our expertise on metric and kernel learning, [2], [3], as well as to explore the use of deep learning methods, [1].

The position is partially funded by a Swiss National Science Foundation international collaboration project as well as other sources for a period of three years. The first year brut salary is 47000CHF (the standard SNSF funding rate for doctorate studens). There will be a possibility to completement the above amount, for that knowledge of French would be a plus.

The successful candidate will join the data mining and machine learning team of the University of Applied Sciences, Western Switzerland, led by Prof. Alexandros Kalousis, and will enroll as a PhD student at the Computer Science department of the University of Geneva within the VIPER group led by Prof. Stephane Marchand-Maillet. Our research explores a number of different issues such as: learning in high dimensional settings, dimensionality reduction and feature selection, learning with structured data (multiple kernel learning), metric and similarity learning, the exploitation of domain knowledge in the learning process. For a more detailed description the interested candidates may take a look at: and the list of publications within there. The greater Geneva lake area is a world-renowned education and research hub, including not only the University of Geneva, but also EPFL, and IDIAP. It offers considerable opportunities for training and exposure to data mining and machine learning, with a number of research teams being active on these and related fields. In addition the selected candidate will have ample opportunities to participate in the main ML and DM conferences.

The ideal candidate will have:

  • A very solid background in a combination of mathematics and computer science. Special areas of interest include: statistical machine learning, statistics, mathematical optimization, mathematical modelling.
  • He or she should have completed, or about to complete, an MSc in the above areas.
  • A very good understanding of machine learning methods and algorithms; project experience in the area will be a considerable plus.
  • Solid expertise in at least one of Matlab or R.
  • Strong programming skills in scripting languages, such as perl, python, etc.
  • Excellent command of English.
  • Team work capacity.

Candidates should send:

  1. A two page CV.
  2. A one page motivation letter explaining why their skills, knowledge and experience make them a particularly suitable candidate for the given position.
  3. The academic transcripts of their studies.
  4. A 500 words research proposal on the project's topic.
  5. The contact details of three referees; do not send reference letters.

to, note that I will be present at ICML 2014 in Beijing and I will be happy to meet potential candidates there.


Application Deadline

Priority will be given to applications send by the 30th of June 2014, however applications will be accepted until the position is filled. The position status will be indicated here

The position will be available from the 1st of September 2014 with a possibility for a later start if necessary.



[1] Qi Y, Oja M, Weston J, Noble WS. A Unified Multitask Architecture for Predicting Local Protein Properties, PLoS ONE 7(3) 2012

[2] Phong Nguyen, Jun Wang and Melanie Hilario and Alexandros Kalousis. Learning heterogeneous similarity measures for hybrid-recommendations in meta-mining, ICDM 2012

[3] Jun Wang, Huyen Do, Adam Woznica and Alexandros Kalousis. Metric learning with multiple kernels, NIPS 2011.