Speaker: Jennifer Newman Title: Clustering for Machine Learning and Two Imaging Problems Abstract: Suffering from data overload, machine learning has adapted in various ways. Instead of feeding deep networks directly with lots of data, where humans need to create appropriate network architecture and hyperparameters, clustering of the data before training can improve the learning phase. In this talk we give a very short introduction to clustering, discussing the K-means clustering algorithm and then the use of graphs for clustering, as spectral clustering has shown promise on large data. Two problems in imaging are discussed - camera identification and forensic vehicle comparison - where clustering may improve solutions.