Madhu Dhiman, a PhD student at IEOR, IIT Bombay, under the supervision of Prof. N. Hemachandra and Prof. Veeraruna Kavitha. She visited INRIA, Sophia Antipolis, France, for the LION research project, working under Prof. Konstantin Avrachenkov.
The research was focused on Markov Decision Processes (MDPs) with intermittent state information. They modeled state transmissions as a Bernoulli lossy process and reformulated the belief-based MDP optimization problem using dual linear programming (LP) in an infinite-dimensional space. The primary objective was to analyze the duality gap in this formulation and its implications for solving MDPs in belief-space representations.
Additionally, she had the opportunity to present one of her works at the "13th Workshop on Performance Evaluation," held from December 2–4, 2024, in Toulouse.
