Non-profit Moroccan AI community (www.atlasia.ma).

Researcher at the Mathematical and Algorithmic Sciences Lab

Deep Learning for Graph and NLP

The bigdata era has driven the recent development of new algorithms and methods, often based on elementary principles allowing to handle large amounts of data. However, these large dimensional data impair the behavior of traditional methods that deserve to be revisited under the eye of more elaborate tools and methods. A better understanding of these methods in the bigdata regime indeed induces possibilities of improvements, thereby leading to the development of more efficient algorithms. The Random Matrix framework provides a powerful tool to understand and analyse the behavior of simple data models (such as the mixture of Gaussians model) in the large dimensional setting, which is naturally the case in the BigData paradigm. My PhD thesis will aim at going beyond the simple models hypothesis, to develop new methods that are more appropriate to practical datasets (structured data, images, etc.).

Consistent sparse principal components estimation using Random Matrix Theory and Concentration of Measure in the high-dimensional setting.

Study and analysis of the ALOHA MAC layer performances in distributed wireless sensor networks (theory, simulations and experimentations).

Non-Gaussian Interference Modelling in Wireless Sensor Networks using Poisson Points Processes.