EE Distinguished Lecturer Seminar: Topological Signal Processing

Thumbnail

Event details

Date 06.06.2019
Hour 16:0017:00
Speaker Prof. Sergio Barbarossa, Sapienza University, Rome
Location
Category Conferences - Seminars
Abstract: The goal of this talk is to present the basic tools for the representation and processing of signals defined over a topological space, i.e. a collection of points only characterized by a set of neighborhood relations. Motivating applications are all signals defined over a non-metric space, like gene regulatory networks, brain networks, social networks, etc. The recent field of graph signal processing (GSP) is a particular case of topological signal processing (TSP), referring to the situation where only pairwise relations among data are taken into account and then focusing only on signals defined over the vertices of a graph. Generalizing GSP, our goal is to incorporate multiway relations of various order by representing signals over simplicial complexes, to exploit their rich algebraic structure. After recalling the basic principles of algebraic topology, we introduce methods to build dictionaries capturing the metric-free structure of the signal domain and leading to informative representations of signals defined over sets of increasing order, e.g., vertices, edges, etc. The identification of these dictionaries forms the basis of a spectral simplicial complex theory, from which we establish an uncertainty principle and present its relation with sampling theory. After having introduced the analysis tools, we consider the synthesis problem, suggesting methods to infer the structure of the simplicial complex from data. We conclude the talk presenting some interesting applications to real data and highlight possible future developments.

Bio: Sergio Barbarossa is a Full Professor at Sapienza University of Rome. He has held several visiting positions at the Environmental Research Institute of Michigan (’88), Univ. of Virginia (’95, ‘97), Univ. of Minnesota (’99). He received the 2010 Technical Achievements Award from the European Association for Signal Processing (EURASIP) society for his contributions on radar, communication and networks and won the IEEE Best Paper Awards from the IEEE Signal Processing Society for the years 2000 and 2014. He is an IEEE Fellow, a EURASIP Fellow and served as an IEEE Distinguished Lecturer. He has been the scientific coordinator of several European projects on wireless sensor networks, small cell networks, and distributed mobile cloud computing. He is currently managing the H2020 EU/Japan project 5G-MiEdge, merging millimeter wave and edge cloud technologies for 5G networks. He is involved in the H2020 project 5G-Conni, for the development of 5G private networks for Industry 4.0. His research interests include signal processing algorithms over topological spaces, topological methods for machine learning, 5G networks and mobile edge computing.

Practical information

  • General public
  • Free

Organizer

  • Prof. Elison Matioli

Contact

Share