Christine Keribin: exploring data to discover and model their structure

To outsiders, nothing seemed to dictate that Christine Keribin would become a lecturer in mathematics. A graduate of the École nationale supérieure des techniques avancées (ENSTA), she began her engineering career at Dassault Systèmes in the 1980s, rising through the ranks to become Head of Development Services. However, over the years, an inner voice constantly reminded her of her appetite for research and sharing knowledge. "I wanted to do a thesis after ENSTA, but those close to me encouraged me to go out into the world of work, which I did," recalls Christine Keribin.

After ten years in industry, she decided on a bold change of direction and began dividing her time between her job as an engineer and a return to studying. "Fascinated by randomness and curious to understand how to model chance, I embarked on a DEA [the equivalent of a 2nd year master’s degree today] followed by a thesis in mathematical statistics," explains the lecturer. After completing her thesis, she spent three years as a part-time associate professor at the Orsay Mathematics Laboratory (LMO - Univ. Paris-Saclay/CNRS) while continuing to work at Dassault Systèmes. In 2002, she left the corporate world for good when she secured a position as a lecturer at the Orsay Technical Institute (IUT). In 2006, she joined the LMO, becoming a professor in 2024.
 

The choice of statistics

Christine Keribin explains her interest in probabilistic modelling as a kind of counterpoint to what, for years, was central to her role in business. "At Dassault Systèmes, I was working on software for computer-aided design (CAD), which was very deterministic, but I found randomness mysterious. I was fascinated by the idea of structuring uncertainty and curious to understand how to model randomness," she says.

Faced with choosing between statistics and probability when she returned to studying, she decided to focus on a DEA in statistical and stochastic modelling, followed by a thesis in mathematical statistics. "What particularly interested me about statistics was their 'inverse problem' aspect; rather than starting from a model and observing the results, we start from the observations and try to understand what model might have generated them," explains Christine Keribin.

Another important factor in her decision to study statistics was the opportunity to explore not only the theory but also the methodology and applications. "I didn't want to confine myself to the abstract. When I obtained theoretical results, I wanted to implement algorithms to see how they worked in a real-world environment," adds the lecturer, whose subsequent career has confirmed that she made the right choice.
 

From theory to real data

While Christine Keribin's thesis initially focused on the highly theoretical aspects of statistics, in particular the mathematical properties of estimators in certain statistical models, her research gradually began to analyse tangible applications. "Some of the meetings I had with other researchers led me to move from theory to explore application data. I also discovered interesting problems in real-life applications, which in turn fed back into theoretical research," explains the lecturer.

One thing led to another, and driven by curiosity, Christine Keribin became interested in investigating data in fields as varied as genomics, functional MRI brain imaging, forest fires and even corporate data. "In particular, I've worked with Stellantis to improve reliability criteria for mechanical parts and with SNCF to predict passenger numbers on Transilien trains. Today, I'm continuing to explore this link between statistical modelling and applications, by working on flow cytometry data for early cancer detection, for example," she explains. A way of coming full circle for this former industry professional.
 

The desire to share... and encourage

Christine Keribin's return to academia has also been driven by her passion for sharing knowledge and her desire to teach. "When I was at Dassault Systèmes, I used to give tutorials at ENSTA," recalls the lecturer. Since then, she has taught subjects ranging from statistics to machine learning and databases, combining them with software to meet the needs of students and the expectations of the professional world. "I've always considered it our mission to support students in their professional integration and help them build their self-confidence," she adds.

It's not surprising, then, that Christine Keribin takes a keen interest in female students and is actively involved in promoting the role of women in science. "It's hard not to feel concerned by this issue when one of our best female students, who has just secured a prestigious internship abroad, has confided in me that she's reluctant to accept for fear of not being up to the job!" Aware of the urgency of tackling stereotypes, Christine Keribin willingly goes beyond her teaching hours to take part in mentoring programmes within the university.
 

Building bridges with the general public

For over ten years, Christine Keribin has been associated with the French Statistical Society (SFdS), where she is currently Vice-President. Her involvement with the SFdS began with an invitation from a colleague, who was then president of this learned society, to take part in a conference he was organising on the use of different statistical software.

Things progressed from there, and Christine Keribin became increasingly involved in the SFdS, organising sessions with business professionals on their use of statistical tools. "I had some fascinating discussions, which made me realise how important it is to popularise and disseminate knowledge of our discipline to a very wide audience. As statistics and machine learning are everywhere in our society, we need to help the general public understand them better," explains Christine Keribin. Another way for the researcher to share her discipline by building bridges between research, business and the general public.