OREGON STATE UNIVERSITY

You are here

Colloquium Series

Once every week while school is in session, EECS invites a distinguished researcher or practitioner in a computer science or electrical and computer engineering-related field to present their ideas and/or work. Talks are generally targeted to electrical engineering and computer science graduate students. This colloquium series is free and open to everyone.

Upcoming Colloquia

Distributed Fault-Tolerant Optimization

Friday, February 12, 2016 -
4:00pm to 4:50pm
KEC 1003 ** Note the different time **

Speaker Information

Nitin Vaidya
Professor
Department of Electrical and Computer Engineering
University of Illinois at Urbana-Champaign

Abstract

Consider a network of agents 1 to N, wherein each agent i has a local convex cost function Fi(x). For instance, the agents may be robots and each robot’s local cost function may represent its cost of moving to location x for a rendezvous with other robots. The objective here is to identify argument x that minimizes the total cost over all the agents. Similar problems arise in the context of machine learning as well where the goal is to determine the optima of a sum of many cost functions. This is a special case of convex optimization that has received significant attention over the past three decades. In particular, many distributed algorithms have been developed to determine the optima without requiring any single agent to be aware of all the cost functions. Our work addresses a fault-tolerant version of the problem, allowing for some of the agents to crash during the computation, or suffer from Byzantine failures (which can result in malicious behavior). For brevity, this talk will focus on the case when up to f agents may suffer Byzantine failures. The ideal goal then is to determine the optima of the sum of local cost functions of just the fault-free agents. It is easy to show that this problem is not solvable unless the local cost functions exhibit some form of redundancy. The talk will consider the irredundant case, and show that despite the failure of f agents, it is possible to determine the optima of a global cost function obtained as the weighted average of the cost functions of just the fault-free agents, wherein all-but-f fault-free agents are assigned non-trivial weights. The talk will provide the intuition behind our results, and discuss some algorithms. Joint work with Lili Su, a Ph.D. candidate at the University of Illinois at Urbana-Champaign.

Speaker Bio

Nitin Vaidya received the Ph.D. from the University of Massachusetts at Amherst. He is a Professor of Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign. He has held visiting positions at Technicolor Paris Lab, TU-Berlin, IIT-Bombay, Microsoft Research, and Sun Microsystems, as well as a faculty position at the Texas A&M University. Nitin Vaidya has co-authored papers that received awards at several conferences, including 2015 SSS, 2007 ACM MobiHoc and 1998 ACM MobiCom. He is a fellow of the IEEE. He presently serves the Chair of the Steering Committee for the ACM PODC conference, and has previously served as Editor-in-Chief for the IEEE Transactions on Mobile Computing, and Editor-in-Chief for ACM SIGMOBILE publication MC2R. More information at http://disc.ece.illinois.edu/.

Action, Gesture and Spoken Command Recognition in Human-Robot Interaction

Monday, February 22, 2016 -
4:00pm to 4:50pm
KEC 1003

Speaker Information

Petros Maragos
Professor
School of E.C.E.
National Technical University of Athens

Abstract

In this talk we will present some advances from our research in the EU project MOBOT which generally aims at the development of an intelligent active mobility assistance robot. We will focus on one of its main goals: to provide multimodal sensory processing capabilities for human action recognition. Specifically, a reliable multimodal information processing and action recognition system needs to be developed, that will detect, analyze and recognize the human user actions based on the captured multimodal sensory signals and with a reasonable level of accuracy and detail for intelligent assistive robotics. One of the main thrusts in the above effort is the development of robust and effective computer vision techniques to achieve the visual processing goals based on multiple cues such as spatiotemporal RGB appearance data as well as depth data from Kinect sensors. Another major challenge is the integration of recognizing specific verbal and gestural commands in the considered human-robot interaction context. In this presentation we summarize advancements in three tasks of the above multimodal processing system for human-robot interaction (HRI): action recognition, gesture recognition and spoken command recognition. Our multi-sensor spoken command recognition system has been developed in the framework of the EU project DIRHA. More information, related papers and current results can be found in http://cvsp.cs.ntua.gr and http://robotics.ntua.gr.

Speaker Bio

Petros Maragos received the Diploma in E.E. from the National Technical University of Athens (NTUA) in 1980 and the M.Sc. and Ph.D. degrees from Georgia Tech, Atlanta, in 1982 and 1985. In 1985, he joined the faculty of the Division of Applied Sciences at Harvard University, where he worked for eight years as professor of electrical engineering affiliated with the Harvard Robotics Lab. In 1993, he joined the faculty of the School of ECE at Georgia Tech. During periods of 1996-98 he had a joint appointment as director of research at the Institute of Language and Speech Processing in Athens. Since 1998, he has been working as a professor at the NTUA School of ECE. He has held a visiting scientist position at MIT LIDS in fall 2012. He is currently the Director of the NTUA Division of Signals, Control and Robotics, and the Director of the Intelligent Robotics and Automation Lab. His research and teaching interests include signal processing, systems theory, pattern recognition, image processing and computer vision, audio and speech/language processing, cognitive systems, and robotics. In the above areas he has published numerous papers, book chapters, and has also co-edited three Springer research books, one on multimodal processing and interaction and two on shape analysis. He has served as: Associate Editor for the IEEE Trans. on ASSP, IEEE Trans. on PAMI, and editorial board member and guest editor for several journals on signal processing, image analysis and vision; co-organizer of several conferences and workshops, including VCIP'92 (GC), ISMM'96 (GC), VLBV'01 (GC), MMSP'07 (GC), ECCV'10 (PC), ECCV’10 Workshop on Sign, Gesture and Activity, EUSIPCO'12 (TC), 2011 & 2014 Dagstuhl Symposia on Shape, 2015 IROS Workshop on Cognitive Mobility Assistance Robots; member of the IEEE committees on DSP, IMDSP and MMSP. He is currently organizing EUSIPCO 2017 (GC). His is the recipient or co-recipient of several awards for his academic work, including a 1983 Sigma Xi best thesis award, a 1987-1992 National Science Foundation Presidential Young Investigator Award, a 1988 IEEE SPS Young Author Best Paper Award, a 1994 IEEE SPS Senior Best Paper Award, the 1995 IEEE W.R.G. Baker Prize Award,the 1996 Pattern Recognition Society's Honorable Mention Award, the EURASIP 2007 Technical Achievement Award for contributions to nonlinear signal, image and speech processing, and the Best Paper Award of the IEEE CVPR-2011 Gesture Recognition Workshop. He was elected a Fellow of IEEE in 1995 and a Fellow of EURASIP in 2010 for his research contributions.

Magnetics + Mechanics + Nanoscale = Electromagnetics Future

Monday, February 29, 2016 -
4:00pm to 4:50pm
KEC 1003

Speaker Information

Greg P. Carman
Professor
Mechanical and Aerospace Engineering Department
UCLA

Abstract

Efficient control of small scale magnetism presents a significant problem for future miniature electromagnetic devices. In most macroscale electromagnetic systems we rely on a discovery made by Oersted 200 years ago where an electrical current through a wire creates a distributed magnetic field. While this concept works well at large scale, it suffers significant problems at volumes below 1 mm3. One approach to control nanoscale magnetic states is spin-transfer torque (STT). However, experimental measurements on STT memory devices indicates that 100 fJ is required to reorient a bit of memory with an energy barrier of about 0.5 aJ, i.e., at 0.0005 percent efficiency. Therefore, new nanoscale approaches are needed for future miniature electromagnetic devices. Recently, researchers have explored strain-mediated multiferroic composites to resolve this problem. For this material class, a voltage-induced strain alters the magnetic anisotropy of the magneto-elastic elements. These strain-mediated multiferroics consists of a piezoelectric material coupled to magneto-elastic elements to transfer electrical energy to magnetic energy through a mechanical transduction. The coupling coefficient (energy transferred) in piezoelectric materials (e.g., lead zirconate titanate, PZT) is approximately 0.8 while the coupling coefficient in magneto-elastic materials (e.g., Tb-Dy-Fe, Terfenol-D) is of similar magnitude, 0.8. Thus, the amount of energy to overcome a 0.5 aJ bit barrier is potentially only 0.8 aJ, or an efficiency of about 60 percent, neglecting line losses. This presentation reviews the motivation, history, and recent progress in nanoscale strain-mediated multiferroics. Research descriptions include analytical and experimental work on strain-mediated multiferroic thin films, single magnetic domain structures, and superparamagnetic particles. The results indicate efficiencies orders of magnitude superior to STT approaches and presents a new approach to control magnetism. Discussions of future research opportunities and novel applications are included.

Speaker Bio

Greg Carman received the Ph.D. degree from Virginia Polytechnic Institute and State University in 1991. He joined the Mechanical and Aerospace Engineering Department at the University of California, Los Angeles, in 1991. He is the director of a new National Science Foundation Engineering Research Center entitled Translational Applications of Nanoscale Multiferroic Materials (TANMS) and is engineering director of the Center for Advanced Surgical and Interventional Technology in the Department of Surgery at UCLA. He is an associate editor for the Journal of Intelligent Material Systems Structures and for Smart Materials and Structures. He received the Northrop Grumman Young Faculty Award in 1995 and three best paper awards from the American Society of Mechanical Engineers (ASME) in 1996, 2001, and 2007. He was elected Fellow of the ASME in 2003 and was awarded the ASME Adaptive Structures and Material Systems Prize honoring his contributions to smart materials and structures in 2004. In 2015 SPIE honored him with the Smart Structures and Materials (SSM) Lifetime Achievement Award. Presently his research interests focus on analytical modeling, fabrication, and testing of multiferroic (magneto-electric) materials and developing devices for medical applications.

Past Colloquia

Sheng Chen
Monday, February 9, 2015 -
4:00pm to 4:50pm
Sheng Chen, Jesse Hostetler and Michael Hilton
Monday, February 2, 2015 -
4:00pm to 4:50pm
Li Hao, Anh Pham and Sherif Abdelwahab
Monday, January 26, 2015 -
4:00pm to 4:50pm
William H. Sanders
Monday, January 12, 2015 -
4:00pm to 4:50pm
Michel Kinsy
Monday, November 24, 2014 -
4:00pm to 4:50pm
Ron Goldman
Monday, November 17, 2014 -
4:00pm to 4:50pm
Steve Brown
Monday, November 10, 2014 -
4:00pm to 4:50pm
Kyoung-Youm Kim
Monday, November 3, 2014 -
4:00pm to 4:50pm

Pages