Geometric Deep Learning in Biology: Representation Learning for Tissue Analysis and Microscopy Applications

Abstract

Advancements in biomedical imaging demand innovative approaches to integrate and interpret complex biological data. The lack of labelled training data however requires the exploitation of context and symmetries inherent to the data to constrain the search space of plausible representations.

This seminar explores the use of representation learning techniques for multimodal image registration, enabling the alignment of diverse imaging modalities to uncover deeper biological insights.

Additionally, we delve into the application of hyperbolic neural networks to embed structured biological knowledge into latent spaces, capturing hierarchical and relational information inherent in tissue organization.



Password for the online link: 930905