Speaker: Andres Contreras (New Mexico State University)
Title: The Classification of Signals: An Overview of Methods, Applications and New Developments
Abstract: The problem of classification of signals according to some pre-established criteria, is a fundamental question arising in many contexts, with applications to engineering, medicine, astrophysics, etc. Different approaches have been put forth to handle and interpret large pieces of information in order to extract patterns and obtain reliable predictions. We will give an overview of existing tools, including convolutional neural networks, optimal mass transport techniques and invariant representations, and comment on their advantages and disadvantages, both from the theoretical and practical points of view. Finally, we will discuss promising lines of research that could be potentially transformative and lead to significant improvements to the state of the art.