Postdoctoral position in Machine Learning and online advertising, University Grenoble Alps - May 2016
-
last edited byPostdoctoral position in Machine Learning and online advertising, University Grenoble Alps
Title: Dynamic collaborative filtering for on-line advertising
Internet advertising has become a major economic challenge for online selling companies that have to optimise their catalogs in real time, in order to propose to users products that fit the best to their interests and preferences [4]. Conventional collaborative filtering engines cannot be applied in this context as there are very few ordinal information, corresponding mainly to past purchases, linking users to the products of their interests [3]. In the other hand relevant information, mainly quantitative, can be obtained from the analyses of user sessions. The particularity of past purchases and the real-value indicators obtained from such analyses is their interdependency and also their nature, as ordinal outputs can be considered as ground truth while real-valued outputs are predicted from user profiling tools and may contain noise.
The main objective of this project is to develop the new generation of recommendation system for online advertisement that can handle both ordinal and quantitative information, associating users to products. To attain this goal, we first propose to establish a theoretical framework for multi-target learning for recommendation systems that takes into account the interdependencies between the two types of outputs, relating users to products, and also their nature [5].
The interdependency between the outputs could be studied using the hierarchical decomposition of products and a statistical tool for empirical processes extending the Hoeffding inequality [2]. Another axis of research would concern the development of new tools for online advertising embedding different objectives which parameters are binded with adapted regularization terms. The overall model needs to be scalable having very low recommendation time.
The Postdoc is a common project with Kelkoo and Best of Media companies which will provide the data necessary for developing the framework and the models. The Postdoc will also be implied in the supervision of a PhD student who is already working on the subject.
For this position we are looking for highly motivated people, with a passion to work in machine learning, information retrieval and the skills to develop algorithms for prediction in real-life applications [1]. We are looking for an inquisitive mind with the curiosity to use a new and challenging technology that requires a rethinking of collaborative filtering with respect to problem constraints in order to achieve a high payoff in terms of speed and efficiency.
We further seek a candidate with the following additional skills:
- Probability and statistics ;
- The ability to analyze, improve and propose new algorithms ;
- Good knowledge of programming languages with a proved experience is a plus.
All materials should be emailed as a single PDF file to Massih-Reza.Amini [ at ] imag.fr with ‘[Application OnLine Advertising]′ in the subject line.
References
[1] W.-S. Chin, Y. Zhuang, Y.-C. Juan, and C.-J. Lin. A learning-rate schedule for stochastic gradient methods to matrix factorization. In Advances in Knowledge Discovery and Data Mining, pages 442–455. Springer, 2015.
[2] Janson, S. Large Deviations for Sums of Partly Dependent Random Variables. Random Structures and Algorithms, 24(3):234–248, 2004.
[3] Y. Koren, R. Bell, and C. Volinsky. Matrix factorization techniques for recommender systems. Computer, (8):30–37, 2009.
[4] Guide to Online Advertising: http://adjuggler.com/docs/AdJuggler_guidetoonlineadv.pdf
[5] Tutorial on Multi-Target Learning: http://www.ngdata.com/icml-2013-tutorial-multi-target-prediction/
