| Title | : | Rich Graph Structures: Algorithms and Applications |
| Speaker | : | Tarun Kumar (IITM) |
| Details | : | Mon, 24 Apr, 2023 4:00 PM @ Google Meet |
| Abstract: | : | In this thesis, we analyze these two rich graph structures, viz. multilayer networks and hypergraphs, and discuss their applications in modeling multi-tissue datasets and collaboration networks, respectively. In particular, we look at the problems of finding node centrality in multilayer networks and clustering and hyperedge prediction in hypergraphs. We introduce MultiCens, a novel centrality framework that can distinguish within- vs. across-layer connectivity to quantify the "influence" of any gene in a tissue on a query set of genes of interest in another tissue. MultiCens enjoys theoretical guarantees on convergence and decomposability, and performs well on synthetic benchmarks. On human multi-tissue datasets, MultiCens predicts known and novel genes linked to hormones. In the next part of this talk, I focus on another network structure, hypergraphs, and explore two problems, hypergraph clustering and hyperedge prediction. First, we provide a generalization of the modularity maximization framework for clustering on hypergraphs. We also propose an iterative technique that provides refinement over the obtained clusters, as shown by our extensive set of experiments. Second, we propose HPRA (Hyperedge Prediction using Resource Allocation) to predict novel hyperedges with a reasonable computation time. The proposed method is tested to predict missing hyperedges as well as future hyperedges using past data, where it outperforms the existing state-of-the-art methods. |
