PhD (M/F) - Optimal policy as a classification problem
- FTC PhD student / Offer for thesis
- 36 mounth
- Doctorate
This offer is available in English version
This offer is open to people with a document recognizing their status as a disabled worker.
Offer at a glance
The Unit
Institut de Recherche en Informatique de Toulouse
Contract Type
FTC PhD student / Offer for thesis
Working hHours
Full Time
Workplace
31071 TOULOUSE
Contract Duration
36 mounth
Date of Hire
01/09/2026
Remuneration
2300 € gross monthly
Apply Application Deadline : 29 May 2026 23:59
Job Description
Thesis Subject
This PhD project proposes to view the search for an optimal reinforcement-learning policy as a classification problem by exploiting the geometric structure of how optimal actions partition the state space. Instead of learning full value functions, the idea is to directly learn the boundaries where two actions become equally good, which define the regions in which each action is optimal. The project begins with a simple two-dimensional, two-action setting to study how these decision boundaries can be learned efficiently, first through threshold-based updates and then through parameterized frontier functions. It then generalizes the approach to higher-dimensional state and action spaces using gradient-based methods and function approximators such as linear models or neural networks. By focusing on learning these boundaries rather than full value functions, the project aims to develop reinforcement-learning algorithms that require less data and converge faster.
Your Work Environment
he position is based at IRIT (Institut de Recherche en Informatique de Toulouse), a major computer science research laboratory hosting several hundred researchers and PhD students. The PhD researcher will join the ASR (Architecture, Systems and Networks) department, whose research areas include computer networks, distributed systems, and machine learning applied to systems.
The project is embedded in a dynamic scientific environment, with opportunities for collaboration with several researchers in the laboratory working on reinforcement learning and networked systems, as well as with the broader Toulouse AI research ecosystem, in particular through the ANITI chair dedicated to reinforcement learning.
Compensation and benefits
Compensation
2300 € gross monthly
Annual leave and RTT
44 jours
Remote Working practice and compensation
Pratique et indemnisation du TT
Transport
Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€
About the offer
| Offer reference | UMR5505-CHLBOU-106 |
|---|---|
| CN Section(s) / Research Area | Information sciences: bases of information technology, calculations, algorithms, representations, uses |
About the CNRS
The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.
Create your alert
Don't miss any opportunity to find the job that's right for you. Register for free and receive new vacancies directly in your mailbox.