R&D Intern - Fairness in Deep Learning - Paris
Salary not listed • 9th arrondissement of Paris, 75009, Paris, France • Internship
Posted 25d ago
Job Location
9th arrondissement of Paris, 75009, Paris, France
Tech Stack
Remote Work Policy
On-site
Employment Type
Internship
Categories
Machine Learning Engineer
About the job
This is a Master's research internship opportunity for 4 to 6 months, starting in Winter 2026-2027, to be conducted at Artefact's research center in Paris. The internship will focus on studying neural networks with the goal of ensuring prediction independence from continuous sensitive variables like age, income, or skin color. The work will involve exploring and implementing existing bias reduction methods for neural networks, establishing robust comparison protocols, and developing new methods tailored for continuous variables. The project aims to address the challenge of controlling dependence between model outputs and continuous sensitive attributes, a dynamic research area influenced by evolving AI regulations.
Responsibilities
- Explore methods for bias reduction in neural networks concerning continuous sensitive variables.
- Implement existing bias reduction techniques, such as regularizations based on dependency measures (e.g., Maximum Correlation, Generalized Disparate Impact).
- Establish a robust protocol for comparing bias reduction methods, considering the trade-off between performance and fairness.
- Design and experiment with new bias reduction methods specifically adapted for continuous variables.
- Utilize open-source libraries like PyTorch and scikit-learn for reproducible and adoptable code.
Requirements
- Master's degree level (BAC+4/BAC+5) in applied mathematics, statistics, or machine learning from a university or engineering school.
- Internship for a gap year or end-of-studies.
- Availability for a 4 to 6 month internship.
- Start date in Winter 2026-2027.