INTERNSHIP Fairness in Image and Video Generation Methods

Télécharger l'offre au format PDF
Le descriptif de l’offre ci-dessous est en Anglais

Niveau de diplôme exigé : Bac + 4 ou équivalent

Fonction : Stagiaire de la recherche

A propos du centre ou de la direction fonctionnelle

The Centre Inria de l’Université de Grenoble groups together almost 600 people in 22 research teams and 7 research support departments.

Staff is present on three campuses in Grenoble, in close collaboration with other research and higher education institutions (Université Grenoble Alpes, CNRS, CEA, INRAE, …), but also with key economic players in the area.

The Centre Inria de l’Université Grenoble Alpe is active in the fields of high-performance computing, verification and embedded systems, modeling of the environment at multiple levels, and data science and artificial intelligence. The center is a top-level scientific institute with an extensive network of international collaborations in Europe and the rest of the world.

Contexte et atouts du poste

Collaboration RobotLearn and the start-up Veeton

Principales activités

Context

Recent advancements in deep generative models, particularly diffusion-based text-to-image and text-to-video models, have opened up new possibilities for creative and practical applications. These methods are increasingly being used for editing tasks such as virtual try-on, image restoration, and content transformation. However, a critical issue lies in their potential to exhibit biases that lead to unfair performance across different demographic groups. For example, these models may not perform equally well for different age groups, genders, or morphologies, resulting in outcomes that may perpetuate stereotypes or exclude certain groups. Addressing this issue is essential for ethical deployment and ensuring inclusivity in generative model applications.

Project Objectives

The aim of this project is to study the biases inherent in image and video generation methods, with a particular focus on editing tasks like virtual try-on. These tasks often reveal disparities in performance due to the underlying data distributions and model architectures. The candidate will analyze potential unfair performance across demographic groups and propose countermeasures to mitigate these biases. By designing fairer editing methods, the project seeks to contribute to the broader goal of responsible AI development.

Potential extension as a PhD position or engineering contract

 

 

 

Compétences

Skills

  • Strong programming skills in Python and PyTorch.
  • Familiarity with computer vision, probability, and deep generative models.
  • Solid understanding of mathematics and statistical bias analysis.
  • Experience with software development tools like GitHub or GitLab.
  • Good communication skills, initiative, and organizational abilities.

Avantages

  • Subsidized meals
  • Partial reimbursement of public transport costs
  • Leave
  • Possibility of teleworking and flexible organization of working hours

Rémunération

 €4.35 per hour of actual presence at 1 January 2024.

About 590€ gross per month (internship allowance)