In the MERLE (Multimodal Effective Representation Learning of Evolution of birds) project, we are interested in the learning of visual and multimodal representation that can help biologists in the phylogenetic classification and modeling evolution of birds. Currently paleontologists use ad hoc features of bird appearance (color, feather, bones, ...) to model evolution and classify clades. We want to explore how deep learning architectures can help to automatically extract relevant features for these biological investigations. We are currently recruiting two master trainees with the following planned missions:
- data pre-processing (e.g. based on the 'birds of the world' database (for biologists))
- study of representation learning architectures (VAE or equivalent) regarding the features extracted and how to constraint them
- fusion of multiple modalities (sound and image)

The ideal candidate will have a master's degree in artificial intelligence/machine learning (or equivalent), previous experience in deep learning, good teamwork and interest in multidisciplinary research.

The interships will start in February 2023 (flexible date) for 5-6 months in LIRIS and LGL laboratories (Lyon, France). The gratification is around 550€/month.

To apply, please send a mail to mathieu.lefort@liris.cnrs.fr and stefan.duffner@liris.cnrs.fr with your CV, cover letter, academic transcript and any other document you deem relevant.