Céline Comte

Change language


My work aims at evaluating and optimizing the performance of large-scale systems using probabilistic methods. Many of these systems have a heterogeneous structure that can be described by a graph: Internet, data centers, organ donation programs, transportation networks, etc. For these systems, I develop control policies to optimize average performance criteria, such as delay or availability, while respecting the operational constraints specific to each system. The abstraction provided by stochastic modeling and graph theory helps me to better understand the impact of decisions at different time scales. Since my arrival at CNRS, I am particularly interested in developing reinforcement learning algorithms adapted to these systems and modeling their performance thanks to Markovian decision processes.


International Publications

French Publications

Ph.D. Thesis


Talks about my research

Reading Groups