Missions

This project aims at developing plastic neural models that capture the insect navigational plasticity and resilience. 36 months (renewable +12 months)

Activités

Simulation of navigating ant-agent models equipped with evolving, plastic neural networks models, which process visual information and drive motor action. These virtual agents will navigate in virtual reconstructions of ants' natural environment, so as to generate feedbacks from motor action to perception. The evolved agents' behaviours will be compared with the observed behaviours of real ants. The evolved neural architecture will be tuned towards and compared with real insect neural circuits. Possibilty to go with the team to the field (in Seville) if desired.

Compétences

- Strong background in neuroevolutionary algorithms and/or NAS.
- Suggested specifics skills: NEAT, Quality Diversity, Recurrent Networks, Plasticity rules.
- Autonomy and ability to develop custom neuroevolutionary approaches
- Use of computing clusters.
- General skills in Python.
- Skills in 3D image rendering (optional)
- First year post-doc are encouraged

Contexte de travail

The position is funded by the ERC-consolidator-grant project RESILI-ANT, led by Dr. Antoine Wystrach, expert in ant navigation neural models. The participant will be hosted within the project team in the University of Toulouse Paul Sabatier, and work closely with the PI Antoine Wystrach, another post-doc and two PhD students performing behavioural studies in the field and in virtual reality, and a research engineer.

Contraintes et risques

Risks linked to work in front of computer screens.