Celine Beji
Welcome to my website!

"I am a person who lives my life based on intention. I don't do anything without intention because intention determines the outcome of your life. It's like cause and effect."

Oprah Winfrey


Celine Beji

-PhD student in machine learning-

Data Science

About me

I'm a PhD student in LAMSADE computer laboratory, affiliated with CNRS, at Dauphine-PSLUniversity. Within the Data Science and Artificial Intelligence group MILES, I am working on Causal Inference, revisited in the light of the latest advances in machine learning.
After obtaining my master's degree in electronics at ENSEA, I completed a master's degree, co-habilitated with ENSAE ParisTech, in statistical information processing, at MIDO departement, of Dauphine-PSL University.

Today, I aspire to participate in research and technological innovation through R&D DeepTech projects, to develop a breakthrough innovation with a high technological content.

More informations: My online resume.


Supervised by Ph. Jamal ATIF and Florian YGER, my research focuses on Causal Inference estimation, from algorithms using standard mathematical models and Deep Learning methods.
Causal Inference, which determines the causal link between a cause and an effect, is developing in multiple fields, such as medicine, marketing and economics. The aim is not only to predict a phenomon, but understand it. For example, in language translation, beyond a dictionary-based translation that is studied in classical machine learning problems, causality identifies the causal link between words to understand a unknown langage no dictionary exists.


Research project



Beji C., Benhamou E., Bon M., Yger F., Atif J.: Estimating individual treatment effects through causal populations identification. In: Esann (2020) PDF . Lien Git .
Beji C., Yger F., Atif J.: Non parametric estimation of causal populationsin via a auto-encoder. (Inprocessing)

DeepTech project

Cancer is one of the major causes of death in the world and 19.3 million new cases have been diagnosed in 2020, including 4.4 million in Europei. Today, multiple treatments exist, but they have the disadvantage of being heavy and expensive. The objective is therefore to predict the most appropriate treatment, according to the specific features of the patients and their pathology.

MyTreatment is a DeepTech project that use advances in Artificial Intelligence to support personalized medicine and prescription of cancer treatments, through Deep Learning algorithms.



Paris Dauphine-PSL University

Python programming


Data analysis

L3 info MIDO

Machine Learning


IT website tools



If you have some questions or need help, please contact me!

Université Paris-Dauphine

Place du Maréchal de Lattre de Tassigny
75016 Paris

Bureau C602