About
Articles by Sébastien
Activity
5K followers
Experience & Education
Volunteer Experience
-
President - Club Francais
Massachusetts Institute of Technology (MIT)
- 1 year 1 month
Arts and Culture
Organizing cultural events to promote French culture and professional events to help MIT French students with their early career choices.
-
Vice President - Point Gamma
Ecole Polytechnique
- 10 months
Arts and Culture
Leading member of a 30 students team, building from the ground up a 7000 people music festival and managing a budget of $350k.
Publications
-
Data Science Foundry for MOOCs
http://dsaa2015.lip6.fr/
In this paper, we present the concept of data science foundry for data from Massive Open Online Courses. In the foundry we present a series of software modules that transform the data into different representations. Ultimately, each online learner is represented using a set of variables that capture his/her online behavior. These variables are captured longitudinally over an interval. Using this representation we then build a predictive analytics stack that is able to predict online learners…
In this paper, we present the concept of data science foundry for data from Massive Open Online Courses. In the foundry we present a series of software modules that transform the data into different representations. Ultimately, each online learner is represented using a set of variables that capture his/her online behavior. These variables are captured longitudinally over an interval. Using this representation we then build a predictive analytics stack that is able to predict online learners behavior as the course progresses in real time. To demonstrate the efficacy of the foundry, we attempt to solve an important prediction problem for Massive Open Online Courses (MOOCs): who is likely to stopout? Across a multitude of courses, with our complex per-student behavioral variables, we achieve a predictive accuracy of 0.7 AUCROC and higher for a one-week-ahead prediction problem. For a two-to-three-weeks-ahead prediction problem, we are able to achieve 0.6 AUCROC. We validate, via transfer learning, that these predictive models can be used in real time. We also demonstrate that we can protect the models using privacy-preserving mechanisms without losing any predictive accuracy.
Other authorsSee publication -
Stability of Transportation Networks Under Adaptive Routing Policies
21st International Symposium on Transportation and Traffic Theory
Growing concerns regarding urban congestion, and the recent explosion of mobile devices able to provide real-time information to traffic users have motivated increasing reliance on real-time route guidance for the online management of traffic networks. However, while the theory of traffic equilibria is very well-known, much fewer results exist on the stability of such equilibria, especially in the context of adaptive routing policy. In this work, we consider the problem of characterizing the…
Growing concerns regarding urban congestion, and the recent explosion of mobile devices able to provide real-time information to traffic users have motivated increasing reliance on real-time route guidance for the online management of traffic networks. However, while the theory of traffic equilibria is very well-known, much fewer results exist on the stability of such equilibria, especially in the context of adaptive routing policy. In this work, we consider the problem of characterizing the stability properties of traffic equilibria in the context of online adaptive route choice induced by GPS-based decision making. We first extend the recent framework of “Markovian Traffic Equilibria” (MTE), in which users update their route choice at each intersection of the road network based on traffic conditions, to the case of non-equilibrium conditions, while preserving consistency with known existence and uniqueness results on MTE. We then exhibit sufficient conditions on the network topology and the latency functions for those MTEs to be stable in the sense of Lyapunov for a single destination problem. For various more restricted classes of network topologies motivated by the observed properties of travel patterns in the Singapore network, under certain assumptions we prove local exponential stability of the MTE, and derive analytical results on the sensitivity of the characteristic time of convergence on network and traffic parameters. The results proposed in this work are illustrated and validated on synthetic toy problems as well as on the full Singapore road network with real demand and traffic data, and the applicability of our results for online road network analysis, pricing and control is discussed.
Other authors -
-
Transfer Learning for Predictive Models in Massive Open Online Courses
Artificial Intelligence in Education
Data recorded while learners are interacting with Massive Open Online Courses (MOOC) platforms provide a unique opportunity to build predictive models that can help anticipate future behaviors and develop interventions. But since most of the useful predictive problems are defined for a real-time framework, using knowledge drawn from the past courses becomes crucial. To address this challenge, we designed a set of processes that take advantage of knowledge from both previous courses and previous…
Data recorded while learners are interacting with Massive Open Online Courses (MOOC) platforms provide a unique opportunity to build predictive models that can help anticipate future behaviors and develop interventions. But since most of the useful predictive problems are defined for a real-time framework, using knowledge drawn from the past courses becomes crucial. To address this challenge, we designed a set of processes that take advantage of knowledge from both previous courses and previous weeks of the same course to make real time predictions on learners behavior. In particular, we evaluate multiple transfer learning methods. In this article, we present our results for the stopout prediction problem (predicting which learners are likely to stop engaging in the course). We believe this paper is a first step towards addressing the need of transferring knowledge across courses.
Other authorsSee publication
Honors & Awards
-
TIME 100 Best Inventions of 2020
TIME Magazine
-
MIT Technology Review - 35 Innovators under 35
MIT Technology Review
https://www.innovatorsunder35.com/the-list/s%C3%A9bastien-boyer/
-
Forbes 30 under 30
Forbes
https://www.forbes.com/30-under-30/2019/#362fa5e863b0
-
Best Machine learning project Award (1st prize)
MIT (Prof. Erik Brynjolfsson, Prof. Sinan Kayhan Aral)
My algorithms to predict hospital readmission using basic information about patients received the first prize (best project award) in the "Machine learning for the Digital economy" contest from Prof. Erik Brynjolfsson after a prestigious jury (including Andrew McAffee and Claudia Perlich) ranked 15 projects.
-
Jury's congratulations for Research internship
Ecole Polyecthnique
Award for outstanding research internship in Applied Mathematics.
-
Oustanding Investment Award
Ecole Polytechnique
Outstanding student who has distinguished himself through his behavior, dedication and commitment to the student body.
Languages
-
French
Native or bilingual proficiency
-
English
Full professional proficiency
Other similar profiles
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content