Mots-Clés

GAN GenIA ecologie oiseaux communication vision par ordinateur texture habitat biodiversité

Description

Postdoctoral Research Position, Montpellier, France

Generative AI for Studying the Influence of Habitats on the Diversification of Bird Color Patterns

Institution: CNRS, University of Montpellier, FRANCE

Location: Centre for Functional and Evolutionary Ecology – CEFE & Montpellier Laboratory for Computer Science, Robotics, and Microelectronics – LIRMM

Duration: 18 Months

Project Overview: Understanding the origins of diversity in color patterns within the animal kingdom is a pivotal research area in evolutionary biology. The visual structure of habitats is known to influence color patterns through selection for camouflage. Recent theoretical developments also suggest that the visual habitat could influence the design of sexual signals through a mechanism named sensory drive. The influence of sensory drive is empirically supported for colors, but not for patterns, and has never been tested on a large scale. This project aims to use generative artificial intelligence to develop a global-scale test of the influence of visual habitats on the evolution of bird plumage patterns and colors.

Objectives:

1. Contribute to the development of a global database of natural habitats of birds

2. Develop a generative AI model that learns to predict bird plumage from images of their habitats

3. Test the hypothesis that color patterns generated from images of a new habitat more closely resemble the color patterns of real birds living in that habitat than the color patterns of birds living in other habitats

4. Develop explainable AI approaches to identify the features involved

This project is carried out in collaboration with Dr. Chris Cooney and Dr. Gavin Thomas from the University of Sheffield. The natural habitat image database is currently being acquired from images in the citizen science database iNaturalist and other sources. The bird database has already been acquired by the Sheffield team.

Qualifications:

• Education: Ph.D. in Computer Science, Computational Biology, or Evolutionary Biology, completed by the time of appointment.
• Experience: Up to 2 years of postdoctoral experience post-Ph.D (researchers with >2years of experience at the time of recruitment are not eligible to apply)
• Technical Skills:
  • Proficiency in deep learning frameworks (e.g., PyTorch), with articles already published in this field.
• Additional Skills:
  • Experience with generative AI.

• • Knowledge of evolutionary biology and animal coloration is highly desirable.
  • Ability to work collaboratively in an interdisciplinary environment.

• Start of the project: May-June 2025.
• About the Team: The project is co-led by Julien Renoult (CNRS, UMR CEFE), Maximilien Servajean (University of Paul Valéry, Montpellier, UMR LIRMM INS2I) and Jérôme Pasquet (University of Paul Valéry, Montpellier, UMR TETIS). The team combines expertise in evolutionary biology, ornithology, and advanced machine learning techniques including generative AI.
• Salary: €3,451 gross per month/€2,773 net per month before taxation.

Application Process: Interested candidates should submit the following documents:

1. Cover Letter: Detailing research experience, interest in the project, and future research goals.
2. Curriculum Vitae (CV): Including a list of publications and relevant projects.
3. References: Contact information for at least two academic references.

Please send your application to julien.renoult@cefe.cnrs.fr by January 15th.