MSCA Doctoral Candidate position

MSCA Doctoral Candidate position (FINALITY DC2) 1 day ago Be among the first 25 applicants Direct message the job poster from Avignon Université Work description: The Laboratory of Informatics of the University of Avignon (LIA), Avignon, France, offers a PhD position on “Delayed Reinforcement Learning with Teams and Games” in the framework of the MSCA Doctoral Network FINALITY (saFe learnINg for LArge scaLe InTerconnected sYstems). This 3 Y doctoral position is conceived to investigate the tradeoffs between delay constraints and information sets for a set of Markovian controllers which act to attain a common goal. The reference application is the one of multiple orchestrators performing resources allocation (RA). Here, the team of controllers behave as partially coordinated reinforcement learning agents. The aim is to design new lightweight Reinforcement Learning (RA)algorithms, able to converge even if single agents have partial view of the system state and limited or delayed control due to physical or structural constraints. The objective of the thesis. The thesis is at the intersection of delayed and multi‑agent reinforcement learning. The presence of delays for agents implementing a policy in a distributed system requires specific tools for the convergence and the adjustment of local policies. The candidate will explore the problems and the trade‑offs related to the presence of material constraints on the available information sets, e.g., in the specific case of delays or partial information shared among the orchestrators performing RA in communication and computing systems. Maintaining consistency in a team of agents in presence of imperfect information shared by individual ones requires developing specific state augmentation and/or approximation techniques. The candidate is expected to develop a class of algorithm for distributed orchestration of resources able to account for partial information of orchestrators. Also, a related challenge is to characterize the possibility of adaptation of the RL solution to changes in the environment. Objectives: overall, the research work will aim at the following specific objectives conceive models for delayed MDPs in the case of a single or multiple agents; provide RL tools for teams of orchestrators performing RA; devise the trade‑offs between delay, information sets and the efficiency of the produced solutions; apply the above methods to the case of constraints and devise transfer learning variants of the problem; Context: The FINALITY DN advances the theoretical computer science curriculum focusing on the mastery of prompt and safe learning techniques for interconnected systems. The 15 FINALITY trainees will develop and integrate innovative methodological tools specialized for AI‑based resource allocation, particularly in the context of large‑scale critical infrastructures for communication and computing. They will develop and validate AI methods that are safe by respecting system boundaries and are prompt in adapting to the environmental changes. Throughout their research training, the FINALITY candidates will prioritize the principles of fairness and computational parsimony of AI methods. Skills/Qualifications a master's degree (or equivalent) in a relevant field such as applied mathematics, theoretical computer science or mathematical engineering. the ideal profile is an applied mathematician willing to perform theoretical AI research curriculum lacking advanced skills in mathematics will not be considered Specific Requirements mathematical modeling and performance evaluation. tools including but not limited to: control theory, game theory, stochastic processes, Markov theory and machine learning (especially Reinforcement learning, Bayesian Optimization and Neural Networks). an interest and skills in computer programming would be an asset for numerical analysis via, e.g., simulation tools. some technological knowledge of computing and telecommunications are welcome but not mandatory. VERY IMPORTANT: please check MSCA Doctoral Networks Eligibility Criteria; in particular, residence in France for more than 12 months in the past 36 months before hiring leads to automatic rejection. Benefits. The recruited researcher will be part of the FINALITY doctoral programme. FINALITY offers a thorough scientific education in the frame of a doctoral training programme and the possibility to participate in specific international training courses, workshops and conferences. The team of doctoral candidates will participate in a European research project with high international visibility. They will have the possibility to perform research visits to internationally renowned research labs in Europe under a prestigious MSCA Fellowship. Practical benefits include: Competitive Salary: The gross monthly salary is € 3 224, subject to tax and social security deductions. Social Security & Pension: Full coverage under French’s social security and pension schemes. Mobility Allowance: Additional monthly allowance for living costs related to relocation. Other Allowances: Family allowance, long‑term leave allowance, and special needs allowance will be available if needed. Professional Development: Access to high‑quality training, networking opportunities, and career development support. To Apply: please, complete the following 2 steps: Mandatory step 1) you should perform the formal candidature on the university site at: https://adum.fr/as/ed/voirproposition.pl?site=adumR&matricule_prop=66520 Optional step 2): you should signal your interest by sending your CV, cover letter, academic marks for both undergraduate and graduate studies, and recommendation letters in a single pdf document to Seniority level Internship Employment type Contract Job function Education and Training Industries Higher Education Referrals increase your chances of interviewing at Avignon Université by 2x We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI. #J-18808-Ljbffr