Events for the 1st week of March
-
Center for Cyber-Physical Systems and Internet of Things and Ming Hsieh Institute for Electrical Engineering Joint Seminar Series on Cyber-Physical Systems
Mon, Feb 27, 2017 @ 02:00 PM - 03:30 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Geir E. Dullerud, Professor, University of Illinois at Urbana-Champaign
Talk Title: Statistical Validation and Principle-Based Simulation of Complex Cyber-Controlled Systems
Abstract: The talk will focus on simulation and a computational approach to verification of the hybrid mathematical models that are formed when combining physics-based models with discrete-transition models, such as those which model software algorithms. Namely, the mathematical models that arise when for instance considering Cyberphysical Systems, or the Internet of Things.
In many game theory and filtering problems it is not possible to analytically obtain solutions for statistical properties of systems under study, and in the first part of the talk we will describe our recent work on numerical approaches to obtaining estimates of these properties, and the application of the techniques developed to particle filtering. Monte Carlo simulation of Markov processes allows the numerical estimation of their statistical properties from an ensemble of sample system paths. We present methods for generating reduced-variance path ensembles for the tau-leaping discrete-time simulation algorithm, which allows mean stochastic process dynamics to be estimated with substantially smaller ensemble sizes. Our methods are based on antithetic and stratified sampling of Poisson random variates, and we provide a combination of analytical proofs and numerical evidence for their performance, which can frequently be a 2-3 orders of magnitude improvement over standard Monte Carlo. Application examples will be discussed.
The second part of the talk will concentrate on system verification, and will present a new verification algorithm for continuous-time stochastic hybrid systems, whose specifications are expressed in metric interval temporal logic (MITL), by deploying a novel model reduction method. By partitioning the state space of the hybrid system and computing the optimal transition rates between partitions, we provide a procedure to both reduce the system to a continuous-time Markov chain, and the associated specification formulas. We prove that the unreduced formulas hold (or do not) if the corresponding reduced formula on the Markov chain is robustly true (or false) under certain perturbations. In addition, a stochastic algorithm to complete the verification has been developed. We have extended the approach of this algorithm, and have developed a direct stochastic algorithm for probabilistically verifying a certain hybrid system class, and applied this technique to an extensive benchmark problem with realistic dynamics.
Biography: Geir E. Dullerud is the W. Grafton and Lillian B. Wilkins Professor in Mechanical Engineering at the University of Illinois at Urbana-Champaign. There he is also a member of the Coordinated Science Laboratory, where he is Director of the Decision and Control Laboratory (21 faculty); he is an Affiliate Professor of both Computer Science, and Electrical and Computer Engineering. He has held visiting positions in Electrical Engineering KTH, Stockholm (2013), and Aeronautics and Astronautics, Stanford University (2005-2006). Earlier he was on faculty in Applied Mathematics at the University of Waterloo (1996-1998), after being a Research Fellow at the California Institute of Technology (1994-1995), in the Control and Dynamical Systems Department. He has published two books: "A Course in Robust Control Theory", Texts in Applied Mathematics, Springer, 2000, and "Control of Uncertain Sampled-data Systems", Birkhauser 1996. His areas of current research interest include convex optimization in control, cyber-physical system security, cooperative robotics, stochastic simulation, and hybrid dynamical systems. In 1999 he received the CAREER Award from the National Science Foundation, and in 2005 the Xerox Faculty Research Award at UIUC. He is a Fellow of both IEEE (2008) and ASME (2011).
Host: Paul Bogdan
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 132
Audiences: Everyone Is Invited
Contact: Estela Lopez
-
Detecting paralinguistic information from speech and language for clinical applications: Algorithms and information limits
Wed, Mar 01, 2017 @ 10:00 AM - 12:00 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Visar Berisha, Arizona State University
Talk Title: Detecting paralinguistic information from speech and language for clinical applications: Algorithms and information limits
Abstract: The ability to share our thoughts and ideas through spoken
communication is fragile. Even the simplest verbal response requires a
complex sequence of events. It requires thinking of the words that best
convey your message; sequencing these words appropriately; and then
sending signals to the muscles required to produce speech. The slightest
damage to the brain areas that orchestrate these events can manifest in
speech and language problems. These disturbances offer a window into
brain functioning. In the first part of this presentation, I will
present an overview of a number of projects where we use interpretable
measures of speech and language production as proxies for cognitive and
motor health. The algorithms behind this work have practical utility in
clinical applications and can help answer basic research questions
related to dysarthric speech production.
In the second part of the talk, I will discuss new results from
non-parametric statistical signal processing that allow us to
characterize the information limits in speech. In contrast to existing
methods based on machine learning, this work provides a framework to
answer fundamental questions such as 'What are the bounds on how well I
can recover a parameter of interest from speech?' or 'How well should an
optimally trained classifier work for a particular application?'
Biography: Visar Berisha is an Assistant Professor at Arizona State
University with a joint appointment in the School of Electrical Computer
and Energy Engineering and the Department of Speech and Hearing Science.
Prior to joining ASU, Berisha was a research scientist at MIT Lincoln
Laboratory and then Principal Research Engineer for a Fortune 500
company. His research interests include speech analytics, statistical
signal processing, and information theory. Much of his recent work spans
all three of these fields to answer basic questions related to the
limits of information in speech. His research has led to many academic
publications, several licensed patents, and a revenue-positive startup
company. Berisha's work has been featured in the Science section of the
New York Times, on National Public Radio, and a number of other national
media outlets.
Host: Shrikanth Narayanan
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 132
Audiences: Everyone Is Invited
Contact: Tanya Acevedo-Lam/EE-Systems
-
MHI CommNetS Seminar
Wed, Mar 01, 2017 @ 02:00 PM - 03:00 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Dr. Arpan Chattopadhyay, USC
Talk Title: Sequential decision algorithms for as-you-go deployment of wireless relay network along a line
Series: CommNetS
Abstract: We are motivated by the need, in some applications, for impromptu or as-you-go deployment of wireless sensor networks. A person walks along a line, starting from a sink node (e.g., a base-station), and proceeds towards a source node (e.g., a sensor) which is at an a priori unknown location. At equally spaced locations, he makes link quality measurements to the previous relay, and deploys relays at some of these locations, with the aim to connect the source to the sink by a multihop wireless path. In this paper, we consider two approaches for impromptu deployment: (i) the deployment agent can only move forward (which we call a pure as-you-go approach), and (ii) the deployment agent can make measurements over several consecutive steps before selecting a placement location among them (the explore-forward approach). We consider a very light traffic regime, and formulate the problem as a Markov decision process, where the trade-off is among the power used by the nodes, the outage probabilities in the links, and the number of relays placed per unit distance. We obtain the structures of the optimal policies for the pure as-you-go approach as well as for the explore-forward approach. We also consider natural heuristic algorithms, for comparison. Numerical examples show that the explore-forward approach significantly outperforms the pure as- you-go approach in terms of network cost. Next, we propose learning algorithms for the explore-forward approach and the pure as-you-go approach, based on single and two timescale Stochastic Approximation, which asymptotically converge to the set of optimal policies, without using any knowledge of the radio propagation model. We demonstrate numerically that the learning algorithms can converge (as deployment progresses) to the set of optimal policies reasonably fast and, hence, can be practical model-free algorithms for deployment over large regions. Finally, we demonstrate the end-to-end traffic carrying capability of such networks via field deployment.
Biography: Arpan Chattopadhyay obtained his B.E. in Electronics and Telecommunication Engineering from Jadavpur University, Kolkata, India in the year 2008, and M.E. and Ph.D in Telecommunication Engineering from Indian Institute of Science, Bangalore in the year 2010 and 2015, respectively. Then he worked as a postdoc in the group DYOGENE of INRIA/ENS Paris. He joined EE department, USC as a postdoc from November 2016. His host is Prof. Urbashi Mitra. His research interests include optimization, learning and control of wireless networks and cyber-physical systems.
Host: Prof. Ashutosh Nayyar
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Annie Yu
-
Big Data, Streaming Graphs, and the Need for Innovations in Architecture
Fri, Mar 03, 2017 @ 10:30 AM - 11:30 AM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Peter Kogge, University of Notre Dame
Talk Title: Big Data, Streaming Graphs, and the Need for Innovations in Architecture
Abstract: This talk will start with some insights gleaned from looking at real-world big data problems and how they are affected by architecture. The Emu migrating thread architecture is then introduced and compared. A general template for integrated big graph batch and streaming analytic processing is developed, and key graph operations, especially streaming, listed. A discussion follows on how the Emu architecture meshes well with such a dual-mode computing template, with some specific emphasis on machine learning functions.
Biography: Peter M. Kogge received his Ph.D. in EE from Stanford in 1973. From 1968 until 1994 he was with IBM's Federal Systems Division, and was appointed an IBM Fellow in 1993. In August, 1994 he joined the University of Notre Dame as first holder of the endowed McCourtney Chair in Computer Science and Engineering. He has served as both Department Chair and Associate Dean for Research, College of Engineering. He is an IEEE Fellow, a Distinguished Visiting Scientist at JPL, and a founder and Chief Scientist of Emu Solutions, Inc. His research interests are in massively parallel computing paradigms, processing in memory, and the relationship between massive non-numeric applications, emerging technology, and computer architectures.
He holds over 40 patents and is author of two books, including the first text on pipelining. His Ph.D. thesis led to the Kogge-Stone adder used in many microprocessors. Other projects included EXECUBE - the world's first multi-core processor and first processor on a DRAM chip, the IBM 3838 Array processor which was for a time the fastest floating point machine marketed by IBM, and the IOP - the world's second multi-threaded parallel processor which flew on every Space Shuttle. In 2008, he led DARPA's Exascale technology study group, which resulted in a widely referenced report on technologies and architectures for exascale computing, and has had key roles on many other HPC programs. His startup, Emu Solutions, has demonstrated the first scalable system that utilizes mobile threads to attack large-scale big data and big graph problems.
Dr. Kogge has received the Daniel Slotnick best paper award (1994), the IEEE Seymour Cray award for high performance computer engineering (2012), the IEEE Charles Babbage award for contributions to the evolution of massively parallel processing architectures (2014), the IEEE Computer Pioneer award (2015), and the Gauss best paper award for high performance computers (2015).
Host: Viktor Prasanna, EEB 200, prasanna@usc.edu
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Gerrielyn Ramos
-
Big Data, Streaming Graphs, and the Need for Innovations in Architecture
Fri, Mar 03, 2017 @ 10:30 AM - 11:30 AM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Peter Kogge, University of Notre Dame
Talk Title: Big Data, Streaming Graphs, and the Need for Innovations in Architecture
Abstract: This talk will start with some insights gleaned from looking at real-world big data problems and how they are affected by architecture. The Emu migrating thread architecture is then introduced and compared. A general template for integrated big graph batch and streaming analytic processing is developed, and key graph operations, especially streaming, listed. A discussion follows on how the Emu architecture meshes well with such a dual-mode computing template, with some specific emphasis on machine learning functions.
Biography: x
Host: Viktor Prasanna, EEB 200, prasanna@usc.edu
Location: 248
Audiences: Everyone Is Invited
Contact: Gerrielyn Ramos
-
Ming Hsieh Institute Seminar Series on Integrated Systems
Fri, Mar 03, 2017 @ 02:30 PM - 04:00 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Dr. Scott Powell, VP Engineering at Jariet Technologies
Talk Title: CMOS Digital Microwave
Host: Profs. Hossein Hashemi, Mike Chen, Dina El-Damak, and Mahta Moghaddam
More Information: MHI Seminar Series IS - Scott Powell.pdf
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 132
Audiences: Everyone Is Invited
Contact: Jenny Lin