Geometric Graph-Based Methods for High Dimensional Data

Explore geometric graph-based methods for segmenting large datasets with graph-based structures in this lecture by Professor Andrea Bertozzi. Delve into the combination of classical nonlinear PDE-based image segmentation techniques with fast linear algebra methods for computing graph Laplacian spectrum information. Learn about algorithms designed to solve semi-supervised and unsupervised graph cut optimization problems. Discover applications in image processing, including image labeling and hyperspectral video segmentation, as well as machine learning and community detection in social networks. Gain insights into modularity optimization posed as a graph total variation minimization problem. The lecture concludes with a Q&A session, providing an opportunity to deepen understanding of these advanced mathematical concepts and their practical applications in data science and image analysis.