Time: 14:30 to 15:30 IST
Invited Talk: AI for Biomedical Imaging: A Graph Representation Perspective.
Senior Researcher, EPFL, Lausanne, Switzerland
Time: 15:30 to 16:30 IST
Invited Talk: Hypergraph Generation and Continuous Product GNNs
Abstract: Geometric deep learning (GDL) is a field of deep learning that adapts neural network techniques to data with complex geometric structures, such as graphs, simplicial complexes, and hypergraphs. This presentation will focus on two areas within GDL: hypergraph generation and continuous graph neural networks. For hypergraph generation, I will introduce a diffusion-based hypergraph generation (HYGENE) method, which operates on the bipartite representation of hypergraphs. HYGENE starts with a single pair of connected nodes and iteratively expands it to form the target hypergraph. At each step, nodes and hyperedges are added in a localized manner using a denoising diffusion process, which allows for the construction of the global structure before refining local details. I show the connection with 3D mesh generation. In the second part of the talk, I will also present continuous product graph neural networks (CITRUS), which are particularly useful for modeling spatiotemporal data, and could be used for image and video processing.
Assistant Professor, Télécom Paris, France
Enhancing ASL Recognition with GCNs and Successive Residual Connections
Time: 16:30 to 16:50 IST
A. Chakraborti (HRI, Prayagraj, India), U. Sarkar (Variable Energy Cyclotron Centre), T. Samanta (Variable Energy Cyclotron Centre), S. Pal (Variable Energy Cyclotron Centre), A. Das (Jadavpur University, Kolkata, India)
Point-GR: Graph Residual Point Cloud Network for 3D Object Classification and Segmentation
Time: 16:50 to 17:10 IST
M. Meraz (Center of Intelligent Robotics Lab,), M. Ansari (Center of Intelligent Robotics Lab), M. Javed (IIIT Allahabad, India), P. Chakraborty (IIIT Allahabad, India),
Efficient Combinatorial Alignment for Improved Graph Generation using GraphVAEs
Time: 17:10 to 17:30 IST
D. Schwartz, Y. Osmanlioglu, D. Bespalov, A. Shokoufandeh (Drexel University, USA)