Performance of tensor-based machine learning methods for large-scale data

30/04/2022 – 01/05/2022 all-day

Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : Institut de Recherche en Informatique de Toulouse
Durée : 5 to 6 months
Contact :
Date limite de publication : 2022-04-30

Contexte :
Several machine learning problems can be addressed by leveraging
tensor methods, especially in unsupervised settings. This approach typically relies on estimating a low-rank tensor model from a noisy dataset, which is usually a challenging task. In general, it is difficult to anticipate the best (or the actual) estimation performance that can be attained. Nevertheless, recent years saw substantial progress in this direction, with many authors studying the attainable performance of estimators of such models under the assumption that the dimensions of the observed data tensor are large. This setting is particularly relevant for large-scale (also known as “big data”) scenarios, where a large number of observations is available.

Sujet :
The primary goal of this internship is to explore the implications of these recent results for some selected practical machine learning problems such as community detection in hyper-graphs, latent variable model estimation and high-order co-clustering. The intern will thus perform computer simulations aimed at understanding the behavior of estimation algorithms in these target problems, whose performance will be confronted to the existing theoretical predictions. New algorithms and strategies for dealing with these problems may be developed based on the the experimental findings. Scientific dissemination of these developments will be encouraged, via publication of papers and/or participation in scientific events.

Please see the attached file for more information.

Profil du candidat :
We look for strongly motivated candidates with a solid background on mathematics and statistics, having good programming skills in scientific computing languages (Python, Matlab, Julia).

Formation et compétences requises :
Optimization theory, linear algebra, probability and statistics. Knowledge/interest in tensors is a strong plus.

Adresse d’emploi :
2 rue Charles Camichel, 31071 Toulouse

Document attaché : 202112231624_sujet-stage.pdf