Description
Date depot: 1 octobre 2020
Titre: Unsupervised local entropy algorithms for generative models of biological sequences
Directeur de thèse:
Martin WEIGT (LCQB)
Encadrant :
Riccardo ZECCHINA (Italie)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
The scope of the project is to explore the accessibility and the generalization performance of high local entropy regions in the loss landscape of unsupervised learning problems, to design algorithms which extend to the unsupervised domain those that have been recently introduced for deep learning, and to apply them to the study of generative models of biological sequences.
Doctorant.e: Meynard Barthélémy