HyperGraph Classes, Robust Fitting and Clustering. Much of AI, particularly within computer vision, relies on robust fitting. More generally, clustering data (i.e., this part of the image relates to a

Description

HyperGraph Classes, Robust Fitting and Clustering. Much of AI, particularly within computer vision, relies on robust fitting. More generally, clustering data (i.e., this part of the image relates to a table top, that part relates to the legs of the table) in a manner that is robust to outliers (data that arises from measurement errors, irrelevant data for the task, or interfering components). A scientific approach tries to understand what makes such tasks hard or easy (to carry out reliably). What characteristics of the data mean that a more simple approach will be successful, or what characteristics mean a more sophisticated approach is required? Indeed, when is the data too noisy to expect any approach to work reliably? This project aims to increase our understanding of these issues.. Scheme: Discovery Projects. Field: 4603 - Computer Vision and Multimedia Computation. Lead: Prof David Suter