Computer Science and Electrical Engineering
University of Maryland, Baltimore County

Metareasoning in Adaptive Systems

Dr. Joshua Jones
University of Maryland, Baltimore County

1:00-2:15pm Friday, 4 March 2011, ITE 227, UMBC

Metareasoning, or reasoning about reasoning, is a process by which a system explicitly accesses (monitors and/or controls) its own reasoning. It is a widely held belief in AI that metareasoning is a cruicial part of human-level intelligence, and it could be considered part of consciousness. In this talk I will avoid such philosophical claims, and instead focus on some more practical applications of metareasoning in software systems that learn and adapt in changing environments. Specifically, I will give an overview of the basic metareasoning architecture and then discuss three systems: Augur, a classification system; GAIA, an adaptive game-playing system; and MCL, a general-purpose metareasoning shell.

Dr. Joshua Jones is a a postdoctoral researcher at the University of Maryland, Baltimore County, where he works on a number of projects with the CORAL and MAPLE labs. These include the development of novel techniques for grammar induction, learning of branching lexicographic preference models, metareasoning in a robotics domain using the Meta-Cognitive Loop (MCL) reasoning system and applications of machine learning methods in medical and financial domains.

Dr. Jones earned his PhD in May 2010 at the Georgia Institute of Technology, as a member of the Design Intelligence Laboratory led by Ashok Goel. His primary research interest is in the area of Artificial Intelligence. He works mostly on learning, which his dissertation viewed as a result of metareasoning — where changes to an agent's knowledge or processes are produced by a deliberative process of agent introspection (self-diagnosis) and self-repair. His dissertation research involved designing both reasoning processes capable of such learning, and knowledge representations capable of supporting those reasoning processes, all of which are implemented in a system called Augur. Augur is also integrated with several statistical machine learning algorithms, and as such, the work is in part intended to establish a bridge between machine learning and knowledge-based approaches to AI.

Update: Slides from Metareasoning in Adaptive Systems, Fri 3/4/2011, UMBC

· directions · talks · events ·

From Pong to Halo: The Making of a Game

Marc Olano, 6:30-8:00pm Thursday, 24 February 2011, UC 310

Sure you know how to play video games, but have you ever wondered what goes into creating one? Gaming is a multi-billion dollar, world-wide industry and there are increasing opportunities for new ideas. With the growing popularity of web-based and downloadable games, it is becoming easier for independent game designers to be successful. Think you’ve got what it takes? Then put down that controller and join us for a behind the scenes look at game development.

The Dessert and Discussion series engages alumni and outstanding UMBC faculty in dialogue about contemporary social issues and research interested in an informal, intimate environment. Light refreshments, dessert and coffee will be provided for each discussion. You may register for one or more of the following discussions below. *Please note that space is limited to 30 participants per discussion.

Solitons, Self-Induced Transparency,
and Modelocking in Quantum Cascade Lasers

Professor Curtis Menyuk
University of Maryland, Baltimore County

1:00-2:15pm Friday, 18 March 2011, ITE 227, UMBC

Standard semiconductor lasers operate in a limited wavelength range, below about 4 microns. Quantum cascade lasers (QCLs) that operate in the mid-IR and far-IR have important applications to medicine, environmental sensing, and national security. While short pulse lasers (~100 fs) are available for standard semiconductor lasers, that is not the case for QCLs. Standard passive modelocking is hard to do in QCLs because of their long coherence times and short gain recovery times. We propose a fundamentally different approach, based on the self-induced-transparency (SIT) effect, that turns these weaknesses into strengths. Solitons, modelocking, and SIT are all reviewed at the beginning of the talk.

Curtis R. Menyuk was born March 26, 1954. He received the B.S. and M.S. degrees from MIT in 1976 and the Ph.D. from UCLA in 1981. He has worked as a research associate at the University of Maryland, College Park and at Science Applications International Corporation in McLean, VA. In 1986 he became an Associate Professor in the Department of Electrical Engineering at the University of Maryland Baltimore County, and he was the founding member of this department. In 1993, he was promoted to Professor. He was on partial leave from UMBC from Fall, 1996 until Fall, 2002. From 1996 – 2001, he worked part-time for the Department of Defense, co-directing the Optical Networking program at the DoD Laboratory for Telecommunications Sciences in Adelphi, MD from 1999 – 2001. In 2001 – 2002, he was Chief Scientist at PhotonEx Corporation. For the last 20 years, his primary research area has been theoretical and computational studies of lasers, nonlinear optics, and fiber optic communications. He has authored or co-authored more than 220 archival journal publications as well as numerous other publications and presentations. He has also edited three books. The equations and algorithms that he and his research group at UMBC have developed to model optical fiber systems are used extensively in the telecommunications and photonics industry. He is a member of the Society for Industrial and Applied Mathematics. He is a fellow of the American Physical Society, the Optical Society of America, and the IEEE. He is a former UMBC Presidential Research Professor.

· directions · talks · events ·

Computer Science and Electrical Engineering
University of Maryland, Baltimore County

A Cognitive Architecture for
Simulating Bodies and Minds

Professor Sergei Nirenburg
University of Maryland, Baltimore County

1:00-2:15pm Friday, 18 February 2011, ITE 227, UMBC

This talk is an overview of a cognitive architecture that supports the creation and deployment of intelligent agents capable of simulating human-like abilities. The agents, have a simulated mind and may also be supplied with a simulated body. These agents are intended to operate as members of multi-agent teams featuring both artificial and human agents. The agent architecture and its underlying knowledge resources and processors are being developed in a sufficiently generic way to support a variety of applications. In this talk we briefly describe the architecture and two proof-of-concept application systems we have developed within it: the Maryland Virtual Patient (MVP) system for training medical personnel and the CLinician’s ADvisor (CLAD).We organize the discussion around four specific aspects of agent capabilities implemented in MVP and CLAD: physiological simulation, knowledge management and learning, decision-making and language processing.

This is joint work with Marjorie McShane and Stephen Beale, with contributions from Jesse English, Ben Johnson, Bryan Wilkinson and Roberta Catizone.

· directions · talks · events ·

Computer Science and Electrical Engineering
University of Maryland, Baltimore County

GPU Computing: Past, Present, and Future

Dr. David Luebke
Director of Research, NVIDIA Corporation

1:00-2:15pm Friday, 4 February 2011, ITE 227

Modern GPUs have outgrown their graphics heritage in many ways to emerge as the world's most successful parallel computing architecture. The GPUs that consumers buy to play video games provide a level of massively parallel computation in a single chip that was once the preserve of supercomputers. The raw computational horsepower of these chips has expanded their reach well beyond graphics. Today's GPUs not only render video game frames, they also accelerate astrophysics, video transcoding, image processing, protein folding, seismic exploration, computational finance, radioastronomy, heart surgery, self-driving cars – the list goes on and on.

When thinking about the future of GPUs it is important to reflect on the past. How did this peripheral grow into a processing powerhouse found everywhere from medical clinics to radiotelescopes to supercomputers? Why the graphics card and not the modem, or the mouse? Have GPUs really outgrown graphics and will they thus evolve into pure HPC processors? (hint: no)

This talk is intended as a sort of "state of the union" for GPU computing. I'll briefly cover the dual heritage of GPUs, both in terms of supercomputing and the evolution of fixed function graphics pipelines. I'll discuss "computational graphics", the evolution of graphics itself into a general-purpose computational problem, and how that impacts GPU design and GPU computing. Finally I'll describe the important problems and research topics facing GPU computing practitioners and researchers.

David Luebke helped found NVIDIA Research in 2006 after eight years on the faculty of the University of Virginia. Luebke received his Ph.D. under Fred Brooks at the University of North Carolina in 1998. His principal research interests are GPU computing and real-time computer graphics. Luebke's honors include the NVIDIA Distinguished Inventor award, the NSF CAREER and DOE Early Career PI awards, and the ACM Symposium on Interactive 3D Graphics "Test of Time Award". Dr. Luebke has co-authored a book, a SIGGRAPH Electronic Theater piece, a major museum exhibit visited by over 110,000 people, and dozens of papers, articles, chapters, and patents.