Ana Bušić

Researcher at Inria Paris Centre and DI ENS

Head of research group: ARGO (Apprentissage, graphes et optimisation distribué) on learning, graphs, and distributed optimization

Contact Information

Inria
48 Rue Barrault
75013 Paris

Office: C108
Phone: +33 1 80 49 43 35

Research Interests

• Optimization, control, and learning for energy networks: integration of renewable resources and demand-response.


• Stochastic operations research:
  
  
  • Dynamic matching models: stability, performance evaluation, optimization and learning.
  
  
  • Stochastic networks: product forms, performance evaluation, admission control...
  
  
• Reinforcement learning: theory and applications.


• (Controlled) Markov processes: model reduction and algorithms for large Markov chains, structure of optimal policies in Markov decision processes, simulation (coupling techniques).


• Local dynamics on graphs: mixing times; probabilistic cellular automata; interacting particle systems.

Selected talks

• Plenary talk "Online stochastic matching". 26ème congrès annuel de la Société Française de Recherche Opérationnelle et d’Aide à la Décision (ROADEF 2025).


• Invited talk "Multi-agent reinforcement learning for wind farm control". 13ème Atelier en Évaluation des Performances. December 2024.


• Invited talk "Multipartite Matching", 7th Workshop on Cognition & Control, Gainesville FL, January 2024.


• Invited talk " Learning Optimal Policies in Cooperative Mean-Field Models with Kullback-Leibler Regularization". Workshop on restless bandits, index policies and applications in reinforcement learning. November 2023.


• Invited talk "Optimal Control in Dynamic Matching Systems". Stochastic Networks, Applied Probability, and Performance (SNAPP) Seminar Series, 2020. Video of the talk.


• Invited talk Distributed demand control in power grids and ODEs for Markov decision processes, Second Conference on the Mathematics of Energy Markets Wolfgang Pauli Institute, Vienna, 4-9 July 2017.


• Invited talk Challenges of renewable power generation: Virtual energy storage from flexible loads, Workshop EDF Lab' : gestion centralisée/décentralisée des systèmes éléctriques. September 16, 2016.


• Invited talk Approximate Optimality with Bounded Regret in Dynamic Matching Models. Real-Time Decision Making Simons Institute, Berkeley, Jun. 27 - Jul. 1, 2016.


• Virtual Energy Storage through Distributed Control of Flexible Loads, CaFFEET 2015

Publications

Ongoing Projects

• Partner lead for Inria Paris of the project AI-NRGY (Distributed AI-based architecture of future energy systems integrating very large amounts of distributed sources) of PEPR TASE (Technologies Avancées des Systèmes Energétiques), within the France 2030 Program (2023 - 2027).


• Inria lead for project: Apprentissage décentralisé et ses applications industrielles, within Joint IFPEN – Inria lab. PhD thesis of Baptiste Corban (2024-2027)


• Member of the Challenge Inria-EDF (2024 - 2027).

Past Projects

• Inria lead for project: Apprentissage par renforcement profond avec contraintes et démonstrations, within Joint IFPEN – Inria lab. PhD thesis of Claire Bizon- Monroc (2021-2024).


• Bilateral research project with VITO, Belgium. Topic: Algorithmic Game and Distributed Learning for Peer-to- Peer Energy Trading. PhD thesis of I. Shilov (2020-2023).


• Bilateral research project with EDF. Topic: demand dispatch of flexible loads (2019-2021).


• PI of ANR JCJC PARI -- Probabilistic Approach for Renewable Energy Integration: Virtual Storage from Flexible Loads (2017 - 2021).


• PI of Inria Associate Team PARIS with University of Florida (2015 - 2017).


• PI of PGMO research project Decentralized control for renewable integration in smart-grids (2015 - 2017).


• Partner lead for Inria Paris for ANR MARMOTE (2013 - 2017).


• PI of ARC OCOQS (2011 - 2012).


• Participant of ANR MAGNUM (2010 - 2014).

Teaching

• Reinforcement learning, Data science & AI for academics (formation continue), PSL Univ., 18h cours


• Reinforcement learning (M2) MASH and MASEF, Dauphine, PSL University, 24h cours


• Foundations of network models (M2) MPRI, 15h cours


• Modèles et algorithmes de réseaux (M1), ENS Paris, 24h cours


• Structures et algorithmes aléatoires (L3), ENS Paris, 24h cours

Past teaching