General information
Offer title : Phd (M/F) in Reinforcement Learning, application to algorithmic audits (H/F)
Reference : UPR8001-GILTRE-004
Number of position : 1
Workplace : TOULOUSE
Date of publication : 07 November 2025
Type of Contract : FTC PhD student / Offer for thesis
Contract Period : 36 months
Start date of the thesis : 1 February 2026
Proportion of work : Full Time
Remuneration : 2200 € gross monthly
Section(s) CN : 06 - Information sciences: bases of information technology, calculations, algorithms, representations, uses
Description of the thesis topic
Exploration of active auditing techniques for large machine learning models, use of reinforcement learning, potential application to recommender systems.
The PhD will mainly investigate the use of reinforcement learning approaches to enable tractable active auditing, by both relaxing guarantees and by adding work assumptions for proposing efficient algorithms.
Work Context
Localised at LAAS, in the context of the PACMAM ANR project: PACMAM aims to enable practical audits of black-box in-vivo models.
To that end, it will address to key challenges: first by devising request-
efficient solutions that requires the auditor a low budget to conduct her
audit. Second, providing solutions that are tractable on modern models.
Finally, by coping with models dynamism, leveraging previous audit results
to improve the following ones.
The position is located in a sector under the protection of scientific and technical potential (PPST), and therefore requires, in accordance with the regulations, that your arrival is authorized by the competent authority of the MESR.
Constraints and risks
N/A