SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for March
-
CS Colloq: Interactive and Intuitive Appearance Design
Mon, Mar 03, 2008 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Interactive and Intuitive Appearance DesignSpeaker: Prof. Fabio Pellacini (Dartmouth)ABSTRACT:
The appearance of objects comes from the interaction of scene lighting
and surface materials, whose careful definition is necessary to achieve
the remarkable sophistication of today's synthetic imagery.
Currently, appearance design is one of the remaining roadblocks for a
ubiquitous use of computer-generated imagery, since slow user feedback
and cumbersome user interfaces make the process significantly time
consuming for expert designers, and beyond the reach of novices.In this talk, I will present our recent results in rendering accurate
lighting for complex environments where we achieve interactivity by
developing new approximation algorithms that can take advantage of
inherent properties of lighting and of today's commodity hardware
architectures. These algorithms completely change the workflow of
artists from an offline to a fully interactive process.
I will also show results from algorithms that build on this
interactivity to support intuitive user interfaces for appearance
design that drastically simplify the time required for designing
appearance.BIO:
Fabio Pellacini is an assistant professor in computer science
at Dartmouth College. His research focuses on algorithms for interactive,
high-quality rendering of complex environments and for artist-friendly
material and lighting design to support more effective content creation.
Prior to joining academia, Pellacini worked at Pixar Animation Studios
on lighting algorithms, where he received credits on various movie
productions.
Pellacini received his Laurea degree in physics from the University of Parma
(Italy), and his M.S. and Ph.D. in computer science from Cornell University.
Pellacini received an NSF CAREER award and a Dartmouth Junior Faculty
Fellowship for his research contributions.Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Autonomous Development of Skill Hierarchies
Tue, Mar 04, 2008 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Autonomous Development of Skill HierarchiesSpeaker: Ozgur Simsek (UMASS)ABSTRACT:
The broad problem I will address in this talk is design of autonomous agents
that are able to efficiently learn how to achieve desired behaviors in large,
complex environments. I will focus on one essential design component: the
ability to form high-level actions, or skills, from available primitives.
Specifically, I will characterize a useful class of skills in terms of general
properties of an agent's interaction with its environment---in contrast to
specific properties of a particular environment---and describe algorithms for
identifying and acquiring such skills autonomously.BIO:
Ozgur Simsek is a Ph.D. candidate in the Computer Science Department of the
University of Massachusetts at Amherst. Her research focuses on machine
learning, artificial intelligence, and network science.Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Information Retrieval for Virtual Worlds
Wed, Mar 05, 2008 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Information Retrieval for Virtual WorldsSpeaker: Dr. Anton Leuski (ICT)ABSTRACT:
Computer simulated virtual worlds have become increasingly important
in recent years. These worlds range from off-line setups where a
single person interacts with a single computer generated character to
massive on-line worlds where tens of thousands of people come
together interacting with each other and numerous virtual characters.
More and more people are using these computer-simulated environments
for education, training, communication, and entertainment. These
worlds are becoming a source for acquiring and polishing real-world
skills. They are also getting used for modeling and analysis of real-world human behavior patterns. Creating effective tools both for
analysis and construction of virtual words is highly important.In this talk I will show how statistical natural language processing
(NLP) techniques can be applied to address this problem. In the first
part of the talk I will discuss how to use NLP approaches such as
language modeling and conditional random fields to build virtual
characters capable of natural language understanding (NLU). I will
describe three different methods for creating NLU subsystems for
virtual characters of different complexities. I will focus my
presentation on a novel text classification algorithm that supports
creation of simple and effective virtual characters. This algorithm
builds on ideas from cross-lingual information retrieval. I will
describe experiments that show that the algorithm outperforms
traditional classification techniques and remains very robust in the
presence of partially correct language input. In the second part of
the talk, I will show how statistical language modelling, text
classification and clustering can be applied to analyze players'
conversations in an online virtual world and how this analysis can be
used to detect interesting player activities, players participating
in those activities, and interaction patterns.BIO:
Dr. Anton Leuski is a Research Scientist at the Institute for Creative
Technologies with the University of Southern California. He holds a Ph.D. in
Computer Science from the University of Massachusetts at Amherst. His
research interests center around interactive information access,
human-computer interaction, and machine learning. Dr. Leuski's recent work has
focused on natural language problems that facilitate dialog between humans and
virtual characters, specifically language understanding and classification,
natural language generation, and activity detection and tracking in massive
collaborative environments.Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Synthesis of Strategies for Noisy and Non-Noisy Multi-Agent Environments
Thu, Mar 06, 2008 @ 01:00 PM - 02:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Synthesis of Strategies for Noisy and Non-Noisy Multi-Agent EnvironmentsSpeaker: Tsz-Chiu Au (UMD)ABSTRACT:
To create new and better agents in multi-agent environments, we may want to
examine the strategies of several existing agents, in order to combine their
best skills. One problem is that in general, we won¡¦t know what those
strategies are; instead, we¡¦ll only have observations of the agents¡¦
interactions with other agents. In this talk, I describe how to take a set of
interaction traces produced by different pairs of players in a two-player
repeated game, and then find the best way to combine them into a composite
strategy. I also describe how to incorporate the composite strategy into an
existing agent, as an enhancement of the agent¡¦s original strategy. In
cross-validated experiments involving 126 agents (most of which written by
students as class projects) for the Iterated Prisoner¡¦s Dilemma, Iterated
Chicken Game, and Iterated Battle of the Sexes, composite strategies produced
from these agents were able to make improvement to the performance of nearly
all of the agents.The speaker will also talk about a technique, Symbolic Noise Detection (SND),
for detecting noise (i.e., mistakes or miscommunications) among agents in
repeated games. The idea behind SND is that if we can build a model of the
other agent's behavior, we can use this model to detect and correct actions
that have been affected by noise. In the 20th Anniversary Iterated Prisoner's
Dilemma competition, the SND agent placed third in the ¡§noise¡¨ category, and
was the best performer among programs that had no ¡§slave¡¨ programs feeding
points to them. I'll discuss how to combine SND with the strategy synthesis
technique in order to produce agents that perform well in noisy, cooperative
environments.BIO:
Tsz Chiu Au is a graduate student at Dept. at Comp. Sci, Univ. of Maryland.
(expected PhD in 2008). He received his B. Eng. degree from Hong Kong Univ. of
Science and Technology.
His research interests lie in AI planning, multi-agent systems and problem
solving by searching. His research accomplishments include his work on coping
with noise in non zero-sum games, synthesis of strategies from interaction
traces and managing volatile data for planning processes in semantic web
service composition.Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: A Theory of Similarity Functions for Learning and Clustering
Thu, Mar 06, 2008 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: A Theory of Similarity Functions for Learning and ClusteringSpeaker: Maria-Florina Balcan (CMU)Abstract:
Machine Learning has become a highly successful discipline with applications in many different areas of Computer Science. A critical advance that has spurred this success has been the development of learning methods using a special type of similarity functions known as kernel functions. These methods have proven very useful in practice for dealing with many different kinds of data and they also have a solid theoretical foundation. However, it was not previously known whether the benefits of kernels can be realized by more general similarity functions. In our work, we develop a theory of learning with similarity functions that positively answers this question. Furthermore, our theory provides a new and much simpler explanation for the effectiveness of kernel methods.Technically speaking, the existing theory of kernel functions requires viewing them as implicit (and often difficult to characterize) mappings in high dimensional spaces. Our alternative framework instead views kernels directly as measures of similarity and it also generalizes the standard theory in important ways. Specifically, our notions of good similarity functions can be described in terms of natural direct properties of the data, with no reference to implicit spaces, and no requirement that the similarity function be positive semi-definite (as in the standard theory).We also show how our framework can be applied to Clustering: i.e., multi-way classification from purely unlabeled data. In particular, using this perspective, we develop a new model that directly addresses the fundamental question of what kind of information a clustering algorithm needs in order to produce a highly accurate partition of the data. Our work provides the first framework for analyzing clustering accuracy without any strong probabilistic assumptions.Biography:
Maria-Florina Balcan is a Ph.D. candidate at Carnegie Mellon University under the supervision of Avrim Blum. She received B.S. and M.S. degrees from the Faculty of Mathematics, University of Bucharest, Romania. Her main research interests are Computational and Statistical Machine Learning, Computational Aspects in Economics and Game Theory, and Algorithms. She is a recipient of the IBM PhD Fellowship.Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Mechanism Design and Analysis Using Simulation-Based Game Models
Mon, Mar 10, 2008 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Mechanism Design and Analysis Using Simulation-Based Game ModelsSpeaker: Yevgeniy Vorobeychik (UMICH)Abstract:
I present a general framework for automated mechanism design on constrained design spaces when the outcomes of strategic interactions between the mechanism designer and participants are specified using a simulation. At the core of the framework lies a black-box stochastic optimization algorithm which guides the selection process of candidate mechanisms. I demonstrate the efficacy of such an approach using a series of applications to two-player design problems. A critical component of mechanism design based on simulations is an algorithm for approximately solving simulation- based games. I present several such algorithms, one of which is provably convergent, and experimentally assess their relative merits.Biography:
Yevgeniy Vorobeychik is a Ph.D. candidate at the University of Michigan AI Laboratory. He has been a fellow in the STIET (Socio-Technical Infrastructure for Electronic Transactions) program for two years and has received honorable mention in the Computer Science & Engineer honors competition for his work on simulation-based mechanism design. His research interests include electronic commerce, game theory, mechanism design, multi-agent systems, and artificial intelligence.Location: Vivian Hall of Engineering (VHE) - 217
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Optimizing Sensing from Water to the Web
Tue, Mar 11, 2008 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Optimizing Sensing from Water to the WebSpeaker: Andreas Krause (CMU)Abstract:
Where should we place sensors to quickly detect contaminations in drinking water distribution networks? Which blogs should we read to learn about the biggest stories on the web? These problems share a fundamental challenge:
How can we obtain the most useful information about the state of the world, at minimum cost?Such sensing, or active learning, problems are typically NP-hard, and were commonly addressed using heuristics without theoretical guarantees about the solution quality. In this talk, I will present algorithms which efficiently find provably near-optimal solutions to large, complex sensing problems. Our algorithms exploit submodularity, an intuitive notion of diminishing returns, common to many sensing problems; the more sensors we have already deployed, the less we learn by placing another sensor. To quantify the uncertainty in our predictions, we use probabilistic models, such as Gaussian Processes. In addition to identifying the most informative sensing locations, our algorithms can handle more challenging settings, where sensors need to be able to reliably communicate over lossy links, where mobile robots are used for collecting data or where solutions need to be robust against adversaries and sensor failures. I will also present results applying our algorithms to several real-world sensing tasks, including environmental monitoring using robotic sensors, activity recognition using a built sensing chair, deciding which blogs to read on the web, and a sensor placement competition.Bio:
Andreas Krause is a Ph.D. Candidate at the Computer Science Department of Carnegie Mellon University. He is a recipient of a Microsoft Research Graduate Fellowship, and his research on sensor placement and information acquisition received awards at several conferences (KDD '07, IPSN '06, ICML'05 and UAI '05). He obtained his Diplom in Computer Science and Mathematics from the Technische Universität Mˆ¢nchen, where his research received the NRW Undergraduate Science Award.Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Rethinking Network Protocol Design for Large Scale Sensor Networks
Thu, Mar 13, 2008 @ 11:00 AM - 12:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Rethinking Network Protocol Design for Large Scale Sensor NetworksSpeaker: Prof. Prasun Sinha (Ohio State)Abstract:
Designing large-scale, unattended and long-lasting sensor networks requires rethinking about fundamental protocol design principles. Existing network protocols for medium access control, routing, and data collection are often based on network structures that are expensive to compute and maintain. Limited energy resources is one of the prime considerations in the design of large-scale sensor networks. We propose a new paradigm of protocol design termed "structure-free design" that relies primarily on dynamic and opportunistic forwarding decisions. It makes use of packet forwarding techniques based on MAC layer anycasting to take advantage of the broadcast nature of wireless channels. This paradigm can be used to design protocols at various layers of the network stack. In particular, I will show how the MAC layer protocol in presence of duty-cycling, and the data aggregation protocol can be designed with provable performance bounds. These protocols have also been shown to provide significant performance improvement using TinyOS implementation on the Kansei testbed at OSU. This work was motivated by observations made during our DARPA/NEST demonstration in 2004 where the world's largest sensor network of 1000 nodes was deployed over a 1.3 km x 0.3 km region for detection, classification and tracking of intruders. Further information on the DARPA project can be found here: http://ceti.cse.ohio-state.edu/exscalBiography:
Prasun Sinha is currently an Assistant Professor in the Department of Computer Science and Engineering at Ohio State University. His interests are in the area of wireless and sensor networking. Prior to joining OSU he worked at Bell Labs, New Jersey for two years. He holds a PhD from UIUC (2001), MS from Michigan State University (1997) and B.Tech from IIT Delhi (1995), all in Computer Science and Engineering. He is a winner of the prestigious NSF CAREER award in 2006. During his graduate studies he won the Ray Ozzie Fellowship (UIUC, 2000), the Mavis Memorial Scholarship (UIUC, 1999), and the Distinguished Academic Achievement Award (MSU, 1997). More information about his research can be found at http://www.cse.ohio-state.edu/~prasun
Location: Henry Salvatori Computer Science Center (SAL) - 222
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Disambiguation of Textual and Web Data
Tue, Mar 18, 2008 @ 11:00 AM - 12:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Disambiguation of Textual and Web DataSpeaker: Dr. Dmitri V. Kalashnikov(UCI)Abstract:
Effectiveness of decision support, data exploration and scientific discovery tools is closely tied to the *quality of data* on which such techniques are applied. It is well recognized that the outcome of the analysis is only as good as the data on which the analysis is performed. Organizations, today, spend a tangible percent of their budgets on information quality tasks (e.g., removing duplicates, correcting errors, filling missing values, etc.) to improve data quality prior to pushing data through the analysis pipeline. Forrester group estimates that the market for data quality will pass the $1 Billion mark by 2008. Given the critical importance of data quality, many efforts, in both industry and academia, have explored systematic approaches to addressing the information quality challenges. Solutions range from approaches addressing specific problems (e.g., address resolution, merging product catalogs) to generic techniques for de-duplication, record linkage, entity resolution, etc. that work across a wide range of domains.This talk focuses primarily on the *Entity Resolution* challenge that arises because objects in the real world are referred to using references or descriptions that are not always unique identifiers of the objects, leading to ambiguity. Such a problem is especially common when multiple data sources are being fused together to create a single unified data warehouse or when data is derived from unstructured sources (e.g., text documents) or semi-structured sources (e.g., HTML Web pages).The talk will summarize our ongoing disambiguation work. Specifically it will cover our general-purpose, domain-independent, disambiguation framework, which we refer to as Graph-based Disambiguation Framework (GDF). GDF is based on the premise that many real-world datasets are relational in nature and contain not only information about entities but also relationships among them, knowledge of which can be used to disambiguate among representations more effectively. The talk will also briefly cover our disambiguation work on creating spatial awareness from raw textual reports and our state of the art algorithms for solving the Web People Search problem.Key Words: Entity Resolution; Web People Search; Spatial Awareness from Text.Biography:
Dmitri V. Kalashnikov received the diploma cum laude in Applied Mathematics and Computer Science from Moscow State University, Russia, in 1999 and the PhD degree in Computer Science from Purdue University in 2003. Currently, he is a researcher at the University of California, Irvine. He has received several scholarships, awards, and honors, including an Intel Fellowship and Intel Scholarship. His current research interests are in the areas of entity resolution & disambiguation, web people search, spatial situational awareness, moving-object databases, spatial databases, and GIS.Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Biomimetic Legged Locomotion and Odor Guided Behavior for Humanitarian Landmine Detectio
Wed, Mar 19, 2008 @ 03:00 PM - 04:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Biomimetic Legged Locomotion and Odor Guided Behavior for Humanitarian Landmine DetectionSpeaker: Dr. Thrishantha Nanayakkara (Harvard) Abstract:
Humanitarian demining is a pre-requisite to the economic revival of many affected countries like Angola, Afghanistan, Mozambique, Kosovo, Cambodia, Sri Lanka etc. My current research efforts make an attempt to find answers to three fundamental questions: 1. How a trace of explosives can be localized using a sensory signal gradient leading to the source 2. How a robotic platform can guide such a sensor under ground contact force constraints (to avoid detonating mines) in an unstructured environment with soft terrain conditions like in an abandoned mine field 3. How a colony of such robots and sensors can adaptively re-organize their collective behaviors to identify the locations of the explosive traces in the most efficient manner. The talk elaborates a project in progress in Sri Lanka in collaboration with the Harvard University to find answers to the above questions. The current focus is on a heterogeneous system of trained rodents, field robots, and human experts to detect landmines in an unstructured forest environment. A salient feature of the proposed system is that each sub-system (robots, animals, and humans) improve their individual capabilities by interacting with each other while performing an area coverage operation in a minefield.Biography:
Thrishantha was born in 1970 in Galle, Sri Lanka. He graduated with a first class honors degree in Electrical Engineering from the University of Moratuwa in 1996. He secured the MSc degree in Electrical Engineering in 1998 and the PhD degree in Systems Control and Robotics from Saga University in 2001. From 2001 ¡V 2003 he was a postdoctoral research fellow at the Department of Biomedical Engineering, School of Medicine, Johns Hopkins University, Baltimore (www.bme.jhu.edu/~reza ). From 2003 to date, he has been in the faculty of the University of Moratuwa, Sri Lanka. He is the principle investigator of the ¡§laboratory for intelligent field robots¡¨ at the Department of Mechanical Engineering (www.mrt.ac.lk/iarc/thrish ). He has been awarded the outstanding researcher award of the University of Moratuwa in 2006. He has published 4 book chapters, 7 international journal papers, and 26 international conference papers. He was the founding general Chair of the International Conference on Information and Automation with technical co-sponsorship of IEEE region 10. At present he is a research fellow in the School of Engineering and Applied Sciences, Harvard University, USA. Keywords: Field robots for landmine detection, animal-robot cooperation, adaptive control, reinforcement based learning, fuzzy and neural network based control, evolutionary optimization.Location: Charles Lee Powell Hall (PHE) - 223
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Strategy Selection for Noisy Empirical Game Models
Fri, Mar 21, 2008 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Strategy Selection for Noisy Empirical Game ModelsSpeaker: Chris Kiekintveld (UMICH)Abstract:
Game theory offers tools for principled analysis of multi-agent systems. However, many potential applications are not amenable to conventional analytic approaches due to the size of the strategy space, payoff uncertainty, and other complications. I will introduce an alternative approach that uses empirical methods (e.g. simulation) as the basis for modeling and reasoning about the game. I illustrate this methodology with an application to the Trading Agent Competition Supply Chain Management game. The use of empirical models raises a number of challenging research questions. Among them is how players should modify their analysis to account for the uncertainty inherent in their observations. The remainder of my talk focuses on this question, evaluating several algorithms for selecting strategies based on noisy empirical game models.Bio:
Chris Kiekintveld is a Ph.D. candidate at the University of Michigan, working with Michael Wellman. His primary research interest is strategic reasoning in multi-agent systems, including both agent design and mechanism design applications. He is an active participant in the Trading Agent Competition as a lead developer for Deep Maize, one of the most successful agents in the supply chain management game.Location: Vivian Hall of Engineering (VHE) - 217
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Coordinating Multiple Moving Objects: From Robots to Microdroplets
Thu, Mar 27, 2008 @ 11:00 AM - 12:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Coordinating Multiple Moving Objects: From Robots to MicrodropletsSpeaker: Prof. Srinivas AkellaABSTRACT:
Coordinating the collision-free motions of multiple moving objects is
a challenging problem, with applications ranging from automotive
workcells to lab-on-a-chip devices. I will first describe our work on
the coordination of multiple robots with dynamics constraints, with
applications in manufacturing cells and UAV coordination. I will then
describe the coordination of microdroplets in digital microfluidic
"lab-on-a-chip" systems. A digital microfluidic system controls
individual droplets of chemicals on an array of electrodes; the
chemical analysis is performed by moving, mixing, and splitting
droplets. This promising new technology can impact applications in
biological research, point-of-care clinical testing, and biochemical
sensing by offering tremendous flexibility and parallelism through
software control. Since the simultaneous coordination of even tens of
droplets on the array is extremely difficult to program manually, we
are developing modular array layouts and network-style droplet routing
algorithms to automatically enable the flexible coordination of
hundreds of droplets. I will discuss our ongoing work in applying
these algorithms to enable versatile digital microfluidic biochips for
problems in biology.BIO:
Srinivas Akella is with the Computer Science department and Center for
Automation Technologies and Systems at Rensselaer Polytechnic
Institute, Troy, New York. He was a Beckman Fellow at the Beckman
Institute for Advanced Science and Technology, University of Illinois
at Urbana-Champaign, before joining RPI. He received his Ph.D. in
Robotics from the School of Computer Science at Carnegie Mellon
University and his B.Tech. from the Indian Institute of Technology,
Madras. He has received the CAREER award from the National Science
Foundation, and was selected as a Rensselaer Faculty Early Research
Career Honoree. His research interests are in developing optimization
and geometric algorithms for applications in robotics, automation,
microsystems, and biotechnology.Location: Ronald Tutor Hall of Engineering (RTH) - 406
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Stabilizing Internet Routing: or, A Story of Heterogeneity
Thu, Mar 27, 2008 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Stabilizing Internet Routing: or, A Story of HeterogeneitySpeaker: Brighten Godfrey (UC Berkeley) Abstract:
A significant cause of the unreliability of end-to-end communications on the Internet is route instability: dynamic changes in routers' selected paths. Instability is becoming even more problematic due to the increasing prevalence of real-time applications and concerns about the scalability of the Internet routing architecture. Yet Route Flap Damping, the main mechanism for combating instability, has introduced unexpected pathologies and reduced availability. This talk describes a more principled approach to stabilizing Internet routing. First, we characterize the design space by identifying general approaches to achieve stability, and giving theoretical bounds on optimal strategies within each approach. Second, I will describe Stable Route Selection (StaRS), a new mechanism which uses flexibility in route selection to improve stability without sacrificing availability. Simulation and experimental results show that StaRS improves stability and end-to- end reliability while deviating only slightly from preferred routes, and closely approaching our theoretical lower bound. These results indicate that StaRS is a promising, easily deployable way to safely stabilize Internet routing. StaRS's stability improvements are enabled by dramatic heterogeneity in route failure patterns. I will present the case that StaRS is an instance of a much more general principle: that heterogeneity --- variation in reliability, processing speed, bandwidth, or other metrics --- should quite often be viewed as an advantage. This thesis is supported by practical and theoretical results in a variety of settings including distributed hash tables, overlay multicast, and job scheduling.Biography:
Brighten Godfrey's research concerns distributed and networked systems, including Internet routing architecture, distributed algorithms, analysis of networks, peer-to-peer systems and overlay networks. He is presently a Ph.D. candidate advised by Ion Stoica at UC Berkeley.
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: CS Colloquia
-
CS Colloq: Intelligent Tutoring for Planning and Reflection
Mon, Mar 31, 2008 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Title: Intelligent Tutoring for Planning and ReflectionSpeaker: Dr. H. Chad Lane (ICT)Abstract:
Computer tutors have made significant progress since they made their first appearance in the late 1960's. The application of artificial intelligence (AI) techniques led to a "second generation" of intelligent tutoring systems (ITSs) that were able to model cognitive skills, give individualized feedback, and track learning with greater precision. In the late 1990's, ITSs emerged that were able to carry out meaningful dialogues with students in order to promote learning. These modern systems leverage advances in the field of natural language processing and dialogue systems. In this talk, I will present my research on building ITSs that use natural language techniques to address the metacognitive skills of planning and reflection. The first part of the talk will focus on a dialogue-based tutoring system for novice programmers that I developed at the University of Pittsburgh that supports basic problem solving and planning skills. In an evaluation, the primary findings were that students who received tutoring from the ITS exhibited an improved ability to compose plans and displayed behaviors suggestive of thinking at greater levels of abstraction than students in a read-only control group. The second part of the talk will provide of an overview of my more recent work conducted at the USC Institute for Creative Technologies. This research effort focuses on the use of explainable AI, natural language generation, and the development of a model of reflective tutoring to support learning of complex skills in game-based learning environments. Here, we consider the nature of learning in loosely-defined domains, serious games, and how tutoring can be used to promote productive learning behaviors.Biography:
Dr. H. Chad Lane is a Research Scientist at the USC Institute for Creative Technologies who specializes in intelligent tutoring systems and cognitive modeling. Since joining the ICT in the fall of 2004, he has focused on issues related to learning in game-based and immersive environments. Chad earned his Ph.D. in Computer Science from the University of Pittsburgh in 2004 under the direction of Dr. Kurt VanLehn. His dissertation focused on the role of dialogue-based intelligent tutoring for the planning and design of computer programs. http://people.ict.usc.edu/~laneLocation: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: CS Colloquia
