Boris Defourny
I completed my PhD in December 2010, after 5 years of exciting research on models and techniques from Operations Research (Stochastic Programming) and Machine Learning (Supervised Learning, Reinforcement Learning). The applications are in sequential decision making under uncertainty and risk-aware decision making. I was working in the Systems and Modeling Research Unit,
University of Liège,
under the supervision of Louis Wehenkel.
The main tools that I used include convex optimization,
combinatorial optimization,
measure and probability theory,
kernel methods,
and Monte Carlo simulation.
I am now with the Princeton Laboratory for Energy Analysis, at the Operations Research and Financial Engineering Department of Princeton University. My Princeton webpage can be accessed here.
Contact Information
You can still contact me through my legacy email:
bdf at montefiore dot ulg dot ac dot be.
Publications
-
Machine Learning Solution Methods for Multistage Stochastic Programming.
B. Defourny.
PhD dissertation, 2010.
Download from Ulg repository (ORBI).
-
Multistage Stochastic Programming: A Scenario Tree Based Approach to Planning under Uncertainty.
B. Defourny, D. Ernst and L. Wehenkel.
Accepted as a contributing chapter to L.E. Sucar, E.F. Morales, and J. Hoey (ed.), Decision Theory Models for Applications in Artificial Intelligence: Concepts and Solutions. To be published by Information Science Publishing. 51 pages (author preprint).
-
Large Margin Classification with the Progressive Hedging Algorithm.
B. Defourny and L. Wehenkel.
OPT 2009: Second NIPS Workshop on Optimization for Machine Learning, Whistler, Canada, 2009.
-
Bounds for Multistage Stochastic Programs using Supervised Learning Strategies.
B. Defourny, D. Ernst and L. Wehenkel.
Stochastic Algorithms: Foundations and Applications. Fifth International Symposium, SAGA 2009. Lecture Notes in Computer Science, vol. 5792, pp. 61-73, Springer, 2009.
Postprint by Springer. Preprint.
-
Supervised Learning of Intra-Daily Recourse Strategies for Generation Management Under Uncertainties.
B. Cornelusse, G. Vignal, B. Defourny and L. Wehenkel.
2009 IEEE Bucharest PowerTech.
-
Probability Density Estimation by Perturbing and Combining Tree Structured Markov Networks.
S. Ammar, P. Leray, B. Defourny and L. Wehenkel.
Symbolic and Quantitative Approaches to Reasoning with Uncertainty - 10th ECSQARU, Lecture Notes in Artificial Intelligence, vol. 5590, pp. 156-167, Springer, 2009.
-
Planning under Uncertainty, Ensembles of Disturbance Trees and Kernelized Discrete Action Spaces.
B. Defourny, D. Ernst and L. Wehenkel.
Proceedings of the IEEE International Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL 2009), pp. 145-152, Nashville TN, 2009.
-
Risk-Aware Decision Making and Dynamic Programming.
B. Defourny, D. Ernst and L. Wehenkel.
Selected for oral presentation at the NIPS-08 Workshop on Model Uncertainty and Risk in Reinforcement Learning, Whistler, Canada, 2008.
-
Lazy Planning under Uncertainty
by Optimizing Decisions on an Ensemble of Incomplete Disturbance Trees.
B. Defourny, D. Ernst and L. Wehenkel.
In Recent Advances in Reinforcement Learning, 8th European Workshop, EWRL'08,
Lecture Notes in Artificial Intelligence, vol. 5323, pp. 1-14, Springer, 2008.
-
High-Dimensional Probability Density Estimation with Randomized
Ensembles of Tree Structured Bayesian Networks.
S. Ammar, P. Leray, B. Defourny and L. Wehenkel.
In Proceedings of the 4th European Workshop on Probabilistic Graphical Models (PGM 2008), Hirtshals, Denmark, 2008.
-
Projecting Generation Decisions Induced by a Stochastic Program on a Family of Supply Curve Functions.
B. Defourny and L. Wehenkel.
In 3rd Carnegie Mellon Conference on the Electricity Industry, Pittsburgh PA, 2007.
Information for Students
Useful resources for the courses "Introduction to Stochastic Processes" and
"Information and Coding Theory" can be found
here (in french).