Expanding the borders of Multimodal Graph Learning with Sheaf Neural Networks

Add to your list(s) Download to your calendar using vCal

• Mar Gonzàlez i Català
• Tuesday 26 November 2024, 17:00-17:30
• Lecture Theatre 2, Computer Laboratory, William Gates Building.

If you have a question about this talk, please contact Pietro Lio.

Multimodal graph learning (MGL) has become an emerging topic due to the prevalence of multimodal graphs (MGs). Numerous types of multimodal data are present in a graph format, forming MGs where nodes represent entities of heterogeneous types and edges indicate connections amongst them. A core challenge in MGL lies in effectively processing and integrating knowledge from multiple modalities while navigating the complexities of graph topology. In this talk, I will review the state-of-the-art approaches in MGL and introduce a novel framework: multimodal sheaf neural networks. By attaching a cellular sheaf to a standard multimodal graph, this framework aims to provide enhanced control over modality fusion, opening new avenues for more robust and interpretable learning.

This talk is part of the Foundation AI series.

This talk is included in these lists:

• Lecture Theatre 2, Computer Laboratory, William Gates Building

Note that ex-directory lists are not shown.