Abstract: Disentangling multicomponent nonstationary signals into coherent AM-FM modes is usually achieved by identifying «loud» time-frequency trajectories where energy is locally maximum. We will here present an alternative perspective that relies on «silent» points, namely spectrogram zeros. Based on the theory of Gaussian analytic functions, a number of results will be presented regarding the distribution of such zeros considered as a point process in the plane, with repulsive properties. The rationale and the implementation of the zeros-based approach for recovering signals embedded in noise will then be discussed, with an application to the extraction and characterization of actual gravitational wave chirps.
Short bio: Patrick Flandrin received the engineer degree from ICPI Lyon, France, in 1978, and the Doct.-Ing. and Docteur d’État degrees from INP Grenoble, France, in 1982 and 1987, respectively. He joined CNRS in 1982, where he is currently Research Director. Since 1991, he has been with the Signals, Systems and Physics Group, within the Physics Department at ENS de Lyon, France. He is currently President of GRETSI, the French Association for Signal and Image Processing. His research interests include mainly nonstationary signal processing (with emphasis on time-frequency and time-scale methods), scaling stochastic processes and complex systems. He published over 250 research papers and authored one monograph in those areas. Dr. Flandrin was awarded the Philip Morris Scientific Prize in Mathematics (1991), the SPIE Wavelet Pioneer Award (2001), the Prix Michel Monpetit from the French Academy of sciences (2001) and the Silver Medal from CNRS (2010). Past Distinguished Lecturer of the IEEE Signal Processing Society (2010-2011), he is a Fellow of the IEEE (2002) and of EURASIP (2009), and he has been elected member of the French Academy of sciences in 2010.
Abstract: Both the number of devices capable of transmitting and receiving information in a wireless manner as well as the number of applications seeking this form of information exchange has exploded over recent times. Heterogeneity exists at all scales from network topology (cellular, pico and femto cells, for example) to the types of devices and applications. To achieve desired Quality-of-Service for all of these systems necessitates the design, optimization and control of networks of unprecedented size and complexity. In this talk, I will suggest the use of signal processing techniques typically applied to image processing-based applications to the design and control of wireless networks. The success of these methods lies in the networks, as represented by graphs, being ``compressible.’’ As examples of how modern signal processing approaches can be employed, we shall develop methods for joint dynamic spectrum sensing and resource allocation for cross-layer network optimization that exploit sparse network dynamics. We also examine the use of graph wavelets and transforms for the parsimonious representation of network cost functions. The proposed technique allows a considerable reduction in the number of observations needed for accurate estimation of the network cost function and further admits a low complexity method by which to design policies for the large-scale network. Numerical results show that anywhere from an order of magnitude less to half of the observations are needed by the new scheme relative to traditional learning schemes to estimate the value function.
Short bio: Urbashi Mitra received the B.S. and the M.S. degrees from the University of California at Berkeley and her Ph.D. from Princeton University. Prior to her PhD studies, she was a Member of Technical Staff at Bellcore. After a six-year stint at the Ohio State University (OSU), she joined the Department of Electrical Engineering at the University of Southern California (USC), Los Angeles, where she is currently a Dean's Professor of Electrical Engineering. She is the inaugural Editor-in-Chief for the IEEE Transactions on Molecular, Biological and Multi-scale Communications. Dr. Mitra is a Distinguished Lecturer for the IEEE Communications Society for 2015-2016. She is a member of the IEEE Information Theory Society's Board of Governors (2002-2007, 2012-2017) and the IEEE Signal Processing Society's Technical Committee on Signal Processing for Communications and Networks (2012-2016). Dr. Mitra is a Fellow of the IEEE. She is the recipient of: a 2016 United Kingdom Royal Academy of Engineering, Distinguished Visiting Fellowship, a 2015 Insight Magazine STEM Diversity Award, 2012 Globecom Signal Processing for Communications Symposium Best Paper Award, 2012 US National Academy of Engineering Lillian Gilbreth Lectureship, USC Center for Excellence in Research Fellowship (2010-2013), the 2009 DCOSS Applications & Systems Best Paper Award, Texas Instruments Visiting Professor (Fall 2002, Rice University), 2001 Okawa Foundation Award, 2000 OSU College of Engineering Lumley Award for Research, 1997 OSU College of Engineering MacQuigg Award for Teaching, and a 1996 National Science Foundation CAREER Award. She has been an Associate Editor for the following IEEE publications: Transactions on Signal Processing (2012--2015), Transactions on Information Theory (2007-2011), Journal of Oceanic Engineering (2006-2011), and Transactions on Communications (1996-2001). She has co-chaired: (technical program) 2014 IEEE International Symposium on Information Theory in Honolulu, HI, 2014 IEEE Information Theory Workshop in Hobart, Tasmania, IEEE 2012 International Conference on Signal Processing and Communications, Bangalore India, and the IEEE Communication Theory Symposium at ICC 2003 in Anchorage, AK; and was the general co-chair for the first ACM Workshop on Underwater Networks at Mobicom 2006, Los Angeles, CA. She served as co-Director of the Communication Sciences Institute at the University of Southern California from 2004-2007. Her research interests are in: wireless communications, communication and sensor networks, biological communication systems, detection and estimation and the interface of communication, sensing and control.
Abstract: This talk summarizes and systematizes the efforts of the machine learning community to deliver universal learning machines using convex optimization algorithms. For many years, it was thought that universality of mapping functions could only be achieved at the expense of non-convex optimization algorithms. We will review three classes of mappers, the Reservoir Computing Machines, the Extreme Learning Machine and the Kernel Adaptive Filters and will contrast them in terms of methods and design challenges.
Short bio: Jose C. Principe (M’83-SM’90-F’00) is a Distinguished Professor of Electrical and Computer Engineering and Biomedical Engineering at the University of Florida where he teaches advanced signal processing, machine learning and artificial neural networks (ANNs) modeling. He is Eckis Professor of ECE and the Founder Director of the University of Florida Computational NeuroEngineering Laboratory (CNEL) www.cnel.ufl.edu. His primary area of interest is processing of time varying signals with adaptive neural models. The CNEL Lab has been studying signal and pattern recognition principles based on information theoretic criteria (entropy and mutual information).
Dr. Principe is an IEEE Fellow. He was the past Chair of the Technical Committee on Neural Networks of the IEEE Signal Processing Society, Past-President of the International Neural Network Society, and Past-Editor in Chief of the IEEE Transactions on Biomedical Engineering. Dr. Principe has more than 800 publications. He directed 93 Ph.D. dissertations and 65 Master theses. He wrote in 2000 an interactive electronic book entitled “Neural and Adaptive Systems” published by John Wiley and Sons and more recently co-authored several books on “Brain Machine Interface Engineering” Morgan and Claypool, “Information Theoretic Learning”, Springer, and “Kernel Adaptive Filtering”, Wiley.
Abstract: Historically some of the most important themes in Signal Processing and Finance have been revolving around the problem of prediction. Much of the initial impetus to the theory of Filtering and Prediction in Signal Processing came from the pioneering work of Wiener and Kolmogorov, giving rise to an impressive body of interdisciplinary research spanning Signal Processing, Artificial Intelligence, Information Theory and Statistics which we have today. Although a considerable part of the Quantitative Finance essentially deals with the same problem, the reported rate of adoption of the powerful Signal Processing tools and techniques in Finance has traditionally been rather low. This is, in part, certainly due to the proprietary nature of the algorithms deployed by the financial institutions, but it can also be attributed to rather challenging statistical properties of the financial timeseries. In this talk we shall present some historical snapshots of the interplay between Signal Processing and Quantitative Finance and point out some potential future directions for exploiting the wealth of Signal Processing techniques and algorithms in Finance.
Short bio: Vladimir Lucic has spent past seventeen years working as Quantitative Finance professional at European, North American and Asian investment Banks. At present he is Managing Director at Barclays, running the global Statistical Modelling and Development team covering electronic trading of Credit and Interest Rate financial products. Prior to this, Vladimir was Global Head of Equities and Investment Strategies Quantitative Analytics at Barclays. Vladimir holds PhD in Probability (Stochastic Filtering) and MSc (Control Theory) from University in Waterloo, Canada. He has published in premier theoretical and practitioners’ journals, and is at present Academic Visitor at Imperial College, London and Adjunct Professor at Dalhousie University, Canada.
Abstract: McLaren Applied Technologies exists at the cutting edge of the McLaren Technology Group, harnessing the culture, advanced technology and predictive analytics that have been honed over 50 years of competition in the world’s most technologically demanding sport where we have pioneered techniques in data logging, control, telemetry, data analysis and simulation.
Formula One is a powerful proving ground for these technologies as the adoption cycle is very short and the team is always pushing for the next development to provide a competitive edge.
We have now extended market reach to deploy these capabilities into new applications that range from drugs trials and optimal drilling of oil wells to condition monitoring on high-value assets such as manufacturing machines and mining trucks and continue to capitalise on this culture of data reliance, analysis and the convergence of real-time data management, predictive analytics and simulation to deliver high performance in health, transport, energy and most recently financial sectors.
In this talk Andy will present an overview of McLaren’s approach, technologies and some example case studies considering their application to the optimisation of both machine and human performance.
Short bio: Andy is the Software Development Director for McLaren Applied Technologies where he heads up the Software, Simulation and Decision Science departments and is responsible for the research and development of a range of technology platforms and solutions that perform batch and stream-based analytics across the market sectors of Motorsport, Automotive, Healthcare, and Public Transport. Andy’s current focus, driven by prior work on national-scale Cyber Security solutions, is on converging processing systems for national-scale data sets with advanced simulation and model execution. Prior to joining McLaren Applied Technologies, Andy has held a variety of senior research, engineering, and architecture roles within FTSE 100 Telecoms and Defence companies and also within Enterprise software start-ups. He holds MEng and Ph.D. degrees in computer science, from Imperial College London, and has held research posts within its Faculty of Engineering.