Time: April 26, 2013
Speaker: Dr. John Wen (RPI)
Title:
Distributed Control Architecture: from Local Optimization to Global Coordination
Abstract:
Past decade has witnessed the explosive growth of data network,processing power,
sensor miniaturization, and controllable devices.Drawing on the ready availability of such technologies,
feedback control is now increasingly applied to highly complex interconnected systems.
Applications include data network, robot swarm, and heating,cooling and lighting systems in large buildings.
These systems involve a large number of interacting subsystems with own decision making criteria and sensing capabilities (considered as agents). Key issues for such systems involve the relationship between local optimization objectives and global systems-wide constraints,communication structure between agents, systems stability and robustness, and overall systems architecture to support heterogeneous sensors and actuators. We consider three types of systems interconnection structures: 1. Indirect Communication: Agents make local decision but respond to the overall systems behavior without explicit communication or coordination between agents. 2. Neighbor Communication: Agents perform individual functions such as seeking and exploration while conforming to the group constraint such as formation
or consensus. 3. Physical Interaction: Agents perform individual functions while maintaining the stable grasp constraint on a jointly held load. In each case, agents have limited information of other agents and the overall system architecture, yet must still coordinate in a stable fashion to achieve the collective objective. The key to such robust behavior lies in convexity (for optimization) and passivity (for stability). In all cases, if appropriate passive outputs
are used for feedback together with gradient type of update, then robust stability towards the global optimum may be achieved. To integrate the sensing and control in such multi-agent systems, we need a versatile low-overhead distributed control and communication software architecture. There are a number of candidates, such as the Robot Operating System (ROS). This talk will present our own open source alternative, Robot Raconteur, which is fully distributed (no master node), object based, easily connected between programs (such as MATLAB) and devices, and offers
important features such as plug-and-play, secured access, and bridge to other
systems such as ROS.
Bio:
John T. Wen received his B.Eng. from McGill University in 1979, M.S. from University of Illinois in 1981, and Ph.D. from Rensselaer Polytechnic Institute in 1985,
all in Electrical Engineering. From 1981-1982, he was a system engineer at Fisher Controls. From 1985-1988, he was a member of technical staff at the Jet
Propulsion Laboratory. Since 1988, he has been with Rensselaer Polytechnic Institute where he is currently a Professor in the Department of Electrical,
Computer, and Systems Engineering with a joint appointment in the Department of Mechanical, Aerospace, and Nuclear Engineering. Since 2005, he has served as
the Director of a New York State sponsored interdisciplinary research center, Center for Automation Technologies and Systems (CATS), with the participation
of over thirty faculty from nine departments. He also served as the Interim Director of the NSF Smart Lighting
Engineering Research Center from June-Dec, 2009. Dr. Wen was an ASEE/NASA Summer Faculty Fellow in 1993, and a Japan Society for the Promotion of Science (JSPS) Senior
Visiting Scientist in 1997. He was an Oversea Assessor for the Chinese Academy of Sciences 2004-2009. He is the
co-inventor of the Adaptive Scanning Optical Microscope (ASOM), which was licensed to Thorlabs and developed as an award-winning product. Dr. Wen's research interest lies in the
modeling and control of dynamical systems with applications to motion control, robot manipulation, opto-mechatronics, thermal management, and active flow control. He
has over 200 refereed publications and six patents. Dr. Wen is a Fellow of IEEE.
Time: April 19, 2013
Speaker: Dr. Ansaf Salleb-Aouissi (Columbia)
Title:
Machine Learning Approaches for Prediction of Preterm Birth
Abstract:
The United States spends over 26 billion dollars per annum on the delivery and care of the 12-13% of infants who are born preterm (Berhman et al. 2007). As preterm birth is a major public health problem with profound implications on society, there would be extreme value in being able to identify women at risk of preterm birth during the course of their pregnancy.
Previous research has largely focused on individual risk factors correlated with preterm birth (e.g. prior preterm birth, race, and infection) and less on combining these factors in a way to understand the complex etiologies of preterm birth. Today, there is no widely tested prediction system that combines well-known factors (Davey et al., 2011) with a good prediction technique to provide actionable decisions within a clinical environment (Mercer et al., 1996).
Our ongoing research addresses this gap by conducting a deeper analysis of the preterm prediction study data collected by the NICHD Maternal Fetal Medicine Units (MFMU) Network, a high-quality data for over 3,000 singleton pregnancies having detailed study visits and biospecimen collection at 24, 26, 28 and 30 weeks gestation. Reports from this dataset used relatively straightforward biostatitistical methodologies such as relative risk assessments to measure associations between risk factors and PTB. These methods include descriptive statistics, Pearson correlation, Fisher's exact tests and linear/logistic regression where risk factors are studied independent of each other.
We conduced experiments on this MFMU dataset using non-linear Support Vector Machines to predict mothers at high risk of PTB at different time points, the main visits in the preterm prediction study. The generality of the models were assessed through cross-validation and then tested on a reserved subset that served as new (unseen) data. Our results demonstrate the superiority of non-linear methods in predicting preterm birth. Besides the fact that no time-dependent prediction was ever used on the MFMU dataset, We obtained an average of sensitivity and specificity in predicting PTB of 56% and 68% respectively, well above the ~21% for sensitivity and ~30% for specificity reported in the literature.
We also provide our initial efforts toward harnessing Electronic Health Records to prepare data for preterm birth prediction. We show our preliminary work on a 5-year snapshot of data for mothers and babies from the New York Presbyterian Hospital EHR systems. We give our initial data preparation and statistics w.r.t. preterm birth diagnostics. We stress the challenges faced in preparing EHR data for Machine Learning ranging from restoring the link between mothers and their babies to diagnostic validation.
We finally show how exciting is this application from a machine learning perspective ranging from multiple labels learning, temporal prediction, to multifactorial and privileged information, aspects we would like to pursue as a continuation of this effort.
Joint work with:
Ilia Vovsha, Ashwath Rajan, Axinia Radeva, Hatim Diab, Ashish Tomar,
CCLS Columbia University
Anita Raja, University of North Carolina (UNC) Charlotte
Ronald Wapner, Department of Obstetrics and gynecology, Columbia University Medical Center
Mary McCord, Medical College//Children's Hospital of Wisconsin
Tara Randis, Department of Neonatology, Columbia University Medical Center
Bio:
Ansaf Salleb-Aouissi joined Columbia University’s Center for Computational Learning Systems as an Associate Research Scientist in 2006 after a Postdoctoral Fellowship at INRIA (France). Her research interests lies in Machine Learning. She has worked on large-scale projects including the power grid. Her current research includes pattern discovery, crowdsourcing and medical informatics. Ansaf has published several peer-reviewed papers in high quality journals, conferences and books including TPAMI, ECML, PKDD, COLT, IJCAI, ECAI and AISTAT. Ansaf received an Engineer degree in Computer Science from the University of Science and Technology Houari Boumediene, Algeria, an M.S. and Ph.D. degrees from University of Orleans (France).
Time: April 12, 2013
Speaker: Dr. Torsten Kroeger (Stanford)
Title:
Highly Reactive Robot Motion Generation and Control
Abstract:
In this talk, I will focus on sensor integration in robotic manipulation control systems, and in particular on the instantaneous planning of motion trajectories in response to unforeseen sensor events. An algorithmic concept that enables instantaneous changes from sensor-guided robot motion control (e.g., force/torque or visual servo control) to trajectory-following motion control and vice versa, will be presented. The resulting class of on-line trajectory generation algorithms serves as an intermediate layer between low-level motion control and high-level sensor-based motion planning. This way, it enables robotic systems to perform a kind of "robotic reflex". Samples and use-cases will accompany the talk in order to provide a comprehensible insight into this interesting and relevant field of robotics.
Bio:
Torsten Kroeger is a lecturer and researcher at Stanford University. He finished his studies in electrical engineering at TU Braunschweig, Germany, in 2002. In 2001, he attended an industrial internship at Lenze Corp. in Atlanta, GA. From 2003 to 2009, he was a research fellow at Institut für Robotik und Prozessinformatik at TU Braunschweig, from which he received his Ph.D. degree in 2009. Since 2006, he has also been working as a consulter for Volkswagen AG, KUKA Roboter GmbH, and Manz Automation AG. He is the founder of Reflexxes GmbH, a start-up company working on the development of the Reflexxes Motion Libraries. In 2010, he joined the robotics research group at Stanford Artificial Intelligence Laboratory, where is now works on instantaneous trajectory generation, autonomous hybrid switched-control of robots, and distributed real-time hard- and software systems. He co-organized GWR 2009 and was the Workshops and Tutorials Chair of ICRA 2011 and the Local Arrangement and Registration Chair of IROS 2011. He received the Heinrich Büssing Award, the GFFT Award, two fellowships of the German Research Association, and he was a finalist of the IEEE/IFR IERA Award and the euRobotics TechTransfer Award.
Time: April 5, 2013
Speaker: Dr. Dejing Dou (Univ. Oregon)
Title:
Use of Ontologies in Semantic Association Mining and Information Extraction
Abstract:
Data mining is the nontrivial extraction of implicit, previously unknown, and potentially useful information from data. The measure of what is meant by “useful” to the user is dependent on the user as well as the domain. Therefore, the role of domain knowledge in the data mining process is essential. Ontologies, such as Semantic Web ontologies, have provided a formal way to represent domain knowledge. However, the use of formal ontologies in data mining is still at a rudimentary level. In this talk, I will introduce our recent research efforts on the use of ontologies in two interesting data mining tasks: semantic association mining, which aims to discover frequent itemsets associated by the indirect connections, and information extraction, which extracts certain types of information from natural language text. I will show the technical detail of our approaches and some interesting results from case studies in various domains. Specially, we have been collaborating with the National Center for Biomedical Ontology at Stanford University to mine semantic associations from electronic health records annotated by medical ontologies. We have been collaborating with the College of Education at the University of Oregon to use ontology-based information extraction to provide grades and meaningful feedbacks for student written summaries.
Bio:
Dejing Dou is an Associate Professor in the Computer and Information Science Department at the University of Oregon and leads the Advanced Integration and Mining (AIM) Lab. Currently he is on one year sabbatical as a Visiting Associate Professor at Stanford Center for Biomedical Informatics Research (BMIR). Dejing Dou received his bachelor degree from Tsinghua University, China in 1996 and his Ph.D. degree from Yale University in 2004. His research areas include ontologies, data integration, data mining, biomedical and health informatics, and the Semantic Web. Dejing Dou has published more than 50 research papers, some of which appear in prestigious conferences and journals like KDD, ICDM, SDM, CIKM, ISWC, JIIS and JoDS. His KDD'07 paper was nominated for the best research paper award. In addition to serving on numerous program committees, Dejing Dou has been invited as panelist by the NSF and NIH several times. He is on the Editorial Board of Journal of Data Semantics. Dejing Dou has received over $3 million PI or co-PI research grants from the NSF and the NIH.
Time: March 22, 2013
Speaker: Dr. Vijay Kumar (UPENN)
Title:
Aerial Robot Swarms
Abstract:
Autonomous micro aerial robots can operate in three-dimensional unstructured
environments, and offer many opportunities for environmental monitoring,
search and rescue, and first response. I will describe the challenges in
developing small, agile robots and our recent work in addressing these challenges.
I will also discuss the deployment of large numbers of aerial robots, focusing
on the control and planning problems with applications to cooperative
manipulation and transport, construction, and exploration and mapping.
Bio:
Vijay Kumar is the UPS Foundation Professor in the School of
Engineering and Applied Science at the University of Pennsylvania, and
on sabbatical leave at White House Office of Science and Technology
Policy where he serves as the assistant director for robotics and cyber
physical systems. He received his Bachelors of Technology from the
Indian Institute of Technology, Kanpur and his Ph.D. from The Ohio State
University in 1987. Dr. Kumar is a Fellow of the ASME and IEEE. He is the
recipient of the 1991 NSF Presidential Young Investigator award, the
1996 Lindback Award for Distinguished Teaching at UPenn, the 1997
Freudenstein Award for significant accomplishments in mechanisms
and robotics, the 2012 ASME Mechanisms and Robotics Award, the 2012
IEEE Robotics and Automation Society Distinguished Service Award, and
a 2012 World Technology Network Award. He has
received numerous best paper awards at top robotics conferences and
has served on the editorial boards of the top IEEE and ASME journals
in robotics and automation. He is also a Distinguished Lecturer in the
IEEE Robotics and Automation Society and an elected member of the
Robotics and Automation Society Administrative Committee (2007-2012).
Time: March 15, 2013
Speaker: Dr. Raghu Machiraju (OSU)
Title:
Using Graphs, Trees, and Matrices to Study Populations of Phenotypes
Abstract:
Phenotypes occur at various scales and it is often necessary to examine populations at each scale either separately or in an integrated manner. There are many benefits to a holistic examination of entire populations of subjects both in translational settings and in direct mechanistic studies often conducted in laboratories. Through such studies of populations, one can determine distinct subtypes in the population, or learn of the putative and/or subtle changes wrought by wholesome changes (including genomic). I will first describe some examples of various studies at the molecular, cellular and the tissue level. Later, I will describe our solutions which often required the non-trivial uses of graphs, trees, and matrices and the use of specific correlation and association metrics. For each example, I will sketch the realized workflow and when possible stress the benefits of learning the underlying models and distributions in the measurements. Similarly, when appropriate I will also emphasize the need for visual interaction with the data.
Bio:
Raghu Machiraju is a Professor of Computer Science and Engineering at The Ohio State University (OSU). he also holds appointments in the OSU Medical Center and the Comprehensive Cancer Center. His research interests include computational biology, visualization and imaging. Raghu has mentored a total of thirteen Doctoral students on a variety of topics in visualization and biomedical imaging. He draws much inspiration from signal processing, and machine learning to accomplish his goals. Much of his work is supported by the National Science Foundation and the National Institutes for Health. Finally, Raghu served the visualization community as General and Paper Co-Chairs of IEEE Visualization Conferences and as a General Co-Chair of the First IEEE Symposium on Biological Data Visualization (BioVis).
Time: March 1, 2013
Speaker: Dr. Francis Quek (VT)
Title:
Embodied Human in a Digital World
Abstract:
Humans do not think like computers. Our minds are ‘designed’ for us to function as embodied beings in the world. This is the basic premise of embodied cognition, and by extension, of embodied interaction. In this talk, I shall highlight two key ideas First, that human language is embodied, and that this concept opens the door to computational access to language that complements speech and natural language processing. I shall show how we can access multimodal language by extracting image-bearing characteristics that inform us of the discourse ‘pulses’ from which language springs. We will also show how the temporal event patterns in language and behavior may be modeled structurally using interactive evolutionary computing approaches.
Second, I will overview a set of projects that show that embodied concepts can inform computational approaches to: 1. Support learning and sensemaking of information through technology ecologies, 2. Organize personal information; 3. Support mathematics instruction for the blind; 4. Design E-readers for the blind; and 5. Simulate large crowds that are made up of multiple social groups and individuals.
Bio:
Francis Quek is currently a Professor of Computer Science at Virginia Tech. He directs the Vision Interfaces and Systems Laboratory at the CHCI. Previously, he has been the Director of the Virginia Tech University-level Center for Human-Computer Interaction, and has been affiliated with Wright State University, the University of Illinois at Chicago, the University of Michigan, and Hewlett-Packard. Francis received both his B.S.E. summa cum laude (1984) and M.S.E. (1984) in electrical engineering from the University of Michigan. He completed his Ph.D. in Computer Science at the same university in 1990. Francis is a member of the IEEE and ACM.
He performs highly interdisciplinary research mainly in embodied interaction, notably related to language and discourse (e.g. multimodal verbal/non-verbal interaction), education (e.g. sensemaking, creative storytelling), special populations (individuals who are blind, children, older adults) and human experience (e.g. affective communication). His other research crosses into medical imaging, computer vision and computer graphics. He has published over 150 peer-reviewed journal and conference articles in human-computer interaction, computer vision, and medical imaging.
Time: February 22, 2013
Speaker: Dr. Wlodek Zadrozny (UNCC)
Title:
After Watson
Abstract:
Almost exactly two years ago “IBM’s supercomputer Watson became possibly the first nonhuman millionaire by besting its human competition on the final day of the Jeopardy challenge”. Watson’s victory generated many other colorful statements, some insightful, but many simply entertaining. The system itself and some of its applications have since been described in numerous technical articles and patents. How does press hype relate to technical reality? Where does Watson stand in relation to our goals of building a machine that understands human language? It might be worth it, after two years, to appraise the state of natural language understanding as discipline of computer science and as a collection of business technologies. With this goal in mind I’ll discuss a collection of Watson innovations, as well as a number of small and big questions that Watson wasn’t supposed answer. These questions represent an opportunity for both scientific breakthroughs and technological innovation.
Bio:
Wlodek Zadrozny spent twenty seven years at the IBM T.J.Watson Research Center. The last five years, until January 2013, working on the Watson/DeepQA natural language processing technology and its applications. The title of this presentation is intended to be ambiguous.
Time: February 15, 2013
Speaker: Dr. Shanzhen Gao (UNCC)
Title:
Open Problems In Discrete Mathematics For Mathematicians And Computer Scientists
Abstract:
We will discuss a collection of open problems in discrete mathematics (graph theory, combinatorics, number theory and set theory), some of which have been studied by some researchers in mathematics and computer science. These problems come from biology, computer science, chemistry, physics or pure mathematics. They are easily stated, does not need much background in discrete mathematics, and could easily be understood and worked on by anyone (even a middle school student) who is eager to think about interesting and unsolved mathematical problems. Some of these problems are quite hard and have been open for a long time. Others are newer unsolved problems. We will also discuss some interesting integer sequences which arise from these problems.
For instance: Let f(m,n,s,t) be the number of (0,1) - matrices of size m×n such that each row has exactly s ones and each column has exactly t ones (sm=nt). The determination of f(m,n,s,t) is an unsolved problem, except for very small s, t.
A (0,1) - matrix is a matrix each of whose elements (entries) is 0 or 1.
Can you find f(3,3,3,3), f(3,3,2,2) and f(16,16,3,3)?
Bio:
Shanzhen Gao is Dr. Ken Chen's Ph.D. student in Computer Science at UNCC. He is a mathematician and got his Ph.D. in mathematics in Florida. He has given six invited talks and twenty general talks on various international conferences in combinatorics, graph theory and number theory. He intends to apply mathematics and computer science to each other.
Time: February 8, 2013
Speaker: Dr. Xiang-Yang Li (IIT)
Title:
Large Scale Wireless Network Systems: Theory, Experience, and Lessons
Abstract:
Wireless sensor networks have been extensively studied from many aspects in the last decade. In this talk, I will talk about our recent theoretical results on the asymptotical behavior of large scale sensor networks, our experiences and lessons in building real operational large scale wireless sensor networking systems. In the first part of the presentation, I will discuss the challenges and the lessons we learned from large scale real sensor system deployments. In the second part of the presentation, I will summarize our results on the asymptotical network capacity of large scale wireless sensor networks. In the past 5 years, collaborated with several schools, we deployed prototype sensor networks in the Yellow Sea (OceanSense 2007-2009), in Tian-Mu Mountain (GreenOrbs 2009-2011), and now we are working towards a 4000 sensor system, CitySee, for urban sensing in WuXi City, China. Currently more than 2200 sensor nodes have been deployed, which covers several square kilometers area. These systems provide us a unique opportunity to observe and understand the behavior of large scale sensor networking systems.
Bio:
Dr. Xiang-Yang Li is a professor at Computer Science Department of IIT. He was an Associate Professor (from 2006 to 2012) and Assistant Professor (from 2000 to 2006) of Computer Science at the Illinois Institute of Technology.
He is recipient of China NSF Outstanding Overseas Young Researcher (B). Dr. Li received MS (2000) and PhD (2001) degree at Department of Computer Science from University of Illinois at Urbana-Champaign. He received a Bachelor degree at Department of Computer Science and a Bachelor degree at Department of Business Management from Tsinghua University, P.R. China, both in 1995.
He published a monograph "Wireless Ad Hoc and Sensor Networks: Theory and Applications". He also co-edited the book "Encyclopedia of Algorithms". The research of Dr. Li has been supported by NSF of USA, RGC of HongKong, and NSF of China.
His research interests include the cyber physical systems, wireless networks, mobile computing, privacy and security, and algorithms. Dr. Li is an editor of several journals, including IEEE Transaction on Parallel and Distributed Systems, IEEE Transaction on Mobile Computing. He served at various capacities (conference chair, TPC chair, or local arrangement chair) in a number of conferences. He has graduated eleven PhD students since 2004. For more information about Prof. XiangYang Li, please check his webpage www.cs.iit.edu/~xli.
Time: February 1, 2013
Speaker: Dr. Maria Gini (UMN)
Title:
Teamwork to leverage synergies among agents in search and rescue
Abstract:
When a large-scale disaster strikes, emergency responders are stretched
thin in an effort to locate and rescue people as quickly as possible.
It is critical for the rescuers to share information effectively under
conditions of time pressure, limited communication, and limited knowledge
of the situation.
We study different types of team structures and compare the performance
of uncoordinated individuals, static teams formed in advance, dynamic
teams where agents can change teams as needs arise, and specialized teams
that admit only specific types of agents based on domain knowledge.
We use the RoboCup Rescue Agent Simulator as a testbed to measure the
performance of our teamwork strategies. We measure the effectiveness
of the teaming strategies using task specific metrics (such as time to
save people, or length of travel), and system level metrics (such as
computation time, and scalability to number of agents). Our results
support the hypothesis that teaming improves performance, and that more
specialized and knowledge-rich teaming arrangements perform better.
Bio:
Maria Gini is a Distinguished Professor of the College of Science
and Engineering at the University of Minnesota. She specializes in the
design of multi-robot and multi-agent systems that are capable of making
intelligent decisions. Such systems range from software agents to robots
that move in unstructured and unknown environments. She is a Fellow of
the AAAI and a Distinguished Scientist of the ACM.
Time: January 25, 2013
Speaker: Dr. Jim Plank (UTK)
Title:
Erasure Coded Storage Systems: Recent Work and A Game Changer
Abstract:
Storage systems have grown to the point where failed components are commonplace and must be anticipated. The general technique to protect data from storage failures is erasure coding, which has had a rich 50+ year history. In this talk, we will present the state of the art with respect to erasure coding, and we will detail two recent research projects. The first is a way to leverage "Streaming" instructions that are available on all modern microprocessors to perform complex erasure coding operations at cache line speeds. This is a "game changer" in the world of erasure coding because it allows storage systems to employ much more complex erasure codes than were previously possible. The second project details techniques to recover from single failures in cloud storage systems with reduced I/O, and therefore better performance.
Bio:
Jim Plank is a professor in the EECS department at the University of Tennessee. He has done research on fault-tolerant computing and storage systems for over 20 years. For the past eight years, his sole focus has been on the design, implementation, and performance of erasure codes in storage systems. He has published numerous papers on the topic, including a very popular tutorial on Reed-Solomon codes and a complete treatment of Minimum Density codes for RAID-6. His open-source libraries for Galois Field arithmetic and for general erasure-coding have been in widespread use by industry and academia.
Time: January 18, 2013
Speaker: Dr. Mario Gerla (UCLA)
Title:
Vehicular Cloud Computing
Abstract:
Mobile Cloud Computing is a new field of research that aims to study mobile agents (people, vehicles, robots) as they interact and collaborate to sense the environment, process the data, propagate the results and more generally share resources. Mobile agents collectively operate as Mobile Clouds enabling environment modeling, content discovery, data collection and other mobile applications in a way that is not possible, or not efficient, with the conventional Internet Cloud alone. In this talk we focus on Vehicles. We review Vehicle Cloud applications ranging from urban sensing to intelligent transportation. We address the cooperation between Vehicular Clouds and the Internet Cloud in the context of a vehicular traffic management application.
Bio:
Dr. Mario Gerla is a Professor in the Computer Science Dept at UCLA. He holds an Engineering degree from Politecnico di Milano, Italy and the Ph.D. degree from UCLA. He became IEEE Fellow in 2002. At UCLA, he was part of the team that developed the early ARPANET protocols under the guidance of Prof. Leonard Kleinrock. He joined the UCLA Faculty in 1976.
At UCLA he has designed network protocols including ad hoc wireless clustering, multicast (ODMRP and CODECast) and Internet transport (TCP Westwood). He has lead the ONR MINUTEMAN project, designing the next generation scalable airborne Internet for tactical and homeland defense scenarios. He is now leading several advanced wireless network projects under Industry and Government funding. His team is developing a Vehicular Testbed for safe navigation, content distribution, urban sensing and intelligent transport. Parallel research activities are wireless medical monitoring using smart phones and cognitive radios in urban environments.
He has served as a Technical Program Committee member of many international conferences, and is active in the organization of conferences and workshops, including MedHocNet and WONS. He serves on the IEEE TON Scientific Advisory Board. He was recently recognized with the annual MILCOM Technical Contribution Award for 2011 and the IEEE Ad Hoc and Sensor Network Society Achievement Award in 2011.
Time: November 30, 2012
Speaker: Anita Raja, Mohammad Hasan, Corie Burke and Daniel Ball (UNC Charlotte)
Title: A Multi-perspective Study of Cooperation in Multiagent Systems
Abstract:
A multiagent system (MAS) consists of multiple interacting intelligent entities within an environment. These entities may be artificial or natural; self-interested or cooperative. We will describe the suite of tools and algorithms we have developed to study agent cooperation in a variety of projects. The first project involves a group of agents that cooperate to track tornadoes and other weather phenomenon in a simulated environment. We will describe our decision-theoretic algorithms for decentralized learning and conflict resolution. The second project describes our work in harnessing game theory and network topological information in a novel way to have agents in a complex network to agree on a social convention. In the third project we design a MAS as a reusable means of bringing the viewpoints of power-limited elements of the population and usually overlooked dimensions of value into public policy decision space. We describe our efforts in building ontologies to represent an agent’s own values as well those of its neighbors as well as non-utilitarian mechanisms to resolve conflicts between agents.
Bio:
Time: November 16, 2012
Speaker: Dr. Moti Yung, Google Research
Title: From Protecting a System to Protecting a Global Ecosystem
Abstract: In this talk I will review the state of the art of protection of the evolving paradigm of a global ecosystem,
and I will further describe how security and privacy mechanisms can be introduced into a general software
development project within the ecosystem. An example of such a project will be described (the ADX
system) which will demonstrate the suggested research and development methodology.
Bio: Dr. Moti Yung is a Research Scientist with Google Inc. and he is also
an Adjunct Senior Research Faculty at the Computer Science Department,
Columbia University. Before that he was a member of IBM Research and a
consultant to leading companies and governments, including RSA
Laboratories. His main research interests are in the areas of
Security, Cryptography and Privacy, as well as in Distributed
Computing Algorithms, and related areas in Computer Science. In the
last 30 years he has been working on, both, central issues in the
scientific foundations and theory, as well as on crucial industrial
solutions, and he has published over 400 works. In 2010 Moti delivered
the annual IACR's Distinguished Lecture in Cryptography.
Time: November 9, 2012
Speaker: Prof. Russel l Taylor, UNC Chapel Hill
Title: Locating, engaging, and building visual analytics systems for scientific collaborators.
Abstract: Scientific Visualization is, by its very nature, a discipline that builds tools to support the work of others. It is therefore critical to have great collaborators with real-world problems to drive us forward. Although visualization contests and online data sets provide opportunities to encounter real-world data, there is no substitute for engaged, enthusiastic collaborators. The best collaborators want you to be engaged in the science, not just drawing pictures or animations. The best collaborators are eager to teach you their domain and their language as you work together. The best collaborators have problems that are challenging enough to push the development of new techniques, tools, and systems. And the best collaborators have more ideas and interests than you can keep up with. During my career, I have been blessed by having the best collaborators. I'll present an approach for locating collaborators through course projects. I'll discuss our recent codification of an approach to engage new collaborators in a way that both supports their science and pushes the state of the art in visualization. I'll also discuss our approaches to building tools (interdisciplinary teams, iterated design, going beyond WIMP interfaces, embracing and extending open-source toolkits). I'll also describe lessons learned in our decades of experience building tools for biomedical scientists within our NIH National Research Resource and our recent experience building tools within our NSF-funded Modeling And Data Analysis Initiative.
Time: November 2, 2012
Speaker: Bob Sizemore and Scott Gerard,
IBM Watson Solutions Development
Title: An inside look at IBM Watson
Abstract:
The Watson artificial intelligence system revolutionized technology when it was introduced in 2011. Learning through interactions and leveraging natural language and hypothesis generation, Watson delivers evidence-based responses to drive strategic outcomes. Perhaps best known for its appearances on the television quiz show Jeopardy!, Watson is now being used as a business tool in the healthcare, financial services and government sectors. This presentation will provide a technical overview of Watson, a discussion of its capabilities, and an inside look at how Watson works. Graduate and doctoral students and faculty in CCI and the Belk College are invited to attend.
Time: October 26, 2012
Speaker: Christopher Hudel, PhD Candidate at UNC Charlotte
Title: Finding Compromised Hosts More Quickly: Optimizing use of the MD5 Hashing Algorithm
Abstract:
The current information security threat landscape now includes attackers whose tools, techniques, and processes (TTPs) are designed to provide for a stealthy infiltration of systems, lateral movement, privilege escalation, and exfiltration of sensitive data (typically for purposes of corporate espionage). Once forensically detected, these same TTPs can act as a 'fingerprint' in locating where the attackers have been elsewhere with a network of host computers.
Formal methods to express these 'fingerprints' are documented as an Indicator of Compromise (IO:C) in a number of languages suited for that purpose such as the OpenIOC framework. IOCs typically include information such as process names, directory and file structures, network configurations, web browser cookies, and the MD5 hash values of individual files that concretely identify a system as one that has been compromised
Our research demonstrates that of all of the indicators, the exhaustive search for MD5 hash values across the entire filesystem consumes the largest amount of time. Given the goal of reducing the time to determine if a system is compromised, what can be done to optimize this specific indicator?
We identify the use of a piece-wise (or partial) MD5 hash value of the first {xx} bytes within each file as a substantially faster method to search for specific MD5-based indicators. Even given the time to root out false positives, our novel approach yields a signficant increase in performance and we believe should be considered as a recommended approach within this domain.
Bio:
Christopher Hudel is a 4th-year UNCC PhD Student working within the LIISP (Laboratory of Information Integration, Security, and Privacy), with Dr. Shehab advising. Christopher is pursuing his degree part-time while serving as the Chief Information Security Officer for SPX, a global manufacturing company with operations in more than 35 countries, and approximately 18,000 employees worldwide. Christopher is responsible for the information security and data protection of the systems and networks that support this enterprise and how to face an ever changing threat landscape. It's these issues of security on a global, multi-national scale to which Christopher is applying his over 17 years of information-security-industry knowledge and practice - addressing challenges that face every modern company, including defending against corporate espionage, which serves as an impetus for this research.
Time: October 19, 2012
Speaker: Erin Carroll, PhD Candidate at UNC Charlotte
Title: Food & Mood: A technological intervention in emotional eating
Abstract:
We eat not just because we are hungry and craving nutrients, but also for a host of emotional and habitual reasons. As part of my work with Dr. Mary Czerwinski at Microsoft Research, we investigate just-in-time interventions for emotional eating. This research involves investigating: emotional eating behaviors; intervention design for emotional eating (how and when to intervene); and automatic emotion detection using physiological sensor data collected in a novel, wearable system. Our results show great potential for just-in-time interventions and that we are able to classify emotions accurately in a mobile system.
Bio:
Erin Carroll is a PhD Candidate at UNC Charlotte, working under Dr. Celine Latulipe in the HCILab. Her dissertation research involves developing metrics for evaluating creativity support tools, including a psychometric tool called the Creativity Support Index and a physiological sensor data approach with electroencephalography (EEG) to detect moments of high creative experience using machine learning. Erin will be defending her dissertation next semester.
Time: October 19, 2012
Speaker: Scott Heggen, PhD Candidate at UNC Charlotte
Title: MAD Science: Increasing Engagement in STEM Education through Participatory Sensing
Abstract:
We introduce the Mobile Application Development for Science (MAD Science) curriculum, which utilizes participatory sensing as a central theme to increase middle school students’ engagement and interest in science and technology. Participatory sensing involves the general public in collecting and sharing information about the surrounding environment through the use of sensing (e.g., camera, GPS, accelerometer) and input capabilities on handheld mobile devices, such as smartphones. We present the results of a pilot offering of the MAD Science curriculum as part of a 10-week after-school program for middle school children. Our results indicate the potential for participatory sensing as a tool for increasing engagement in technology; after participating in the MAD Science program, students viewed technology more favorably, indicated increased enjoyment of technology, and indicated increased interest in pursuing education and careers in science and computing.
Bio:
Scott Heggen is a fourth year Ph.D. student at UNCC, seeking a degree in Computing and Information Systems, and is researching methods to improve participatory sensing. He is developing tools to make participatory sensing more attractive and easier to deploy by non-technical persons, and is also interested in methods for improving STEM education through the use of participatory sensing as an informal science education tool.
Time: October 12, 2012
Speaker: Dr. Douglas H. Fisher, Vanderbilt University
Title: Research and Education in Computational Sustainability
Abstract:
Computational sustainability is a growing interdisciplinary field concerned with the application of computational models and methods to problems of environmental and societal sustainability. Computing is needed to grapple with the complexities and uncertainties of even the simplest sustainability challenges. Moreover, computational abstractions shed light on similarities and synergies across sustainability problems and domains. Inversely, sustainability problems motivate new computational methods and abstractions. The talk will overview computational sustainability research directions, describe initiatives that facilitate the integration of research and education in the area, and survey selected history and funding opportunities.
Bio:
Doug Fisher is an associate professor of computer science at Vanderbilt University. He is a faculty member in residence at Vanderbilt, living in McGill Hall with his wife, and active in residential life across campus. He is the co-chair for the Computational Sustainability track at AAAI-13, and area chair for AI and Sustainability of IEEE Intelligent Systems. Doug served as a Program Director at the National Science Foundation (2007-2010), where he oversaw research in artificial intelligence and machine learning. Doug received NSF Director’s Award in 2010 for Excellence “in recognition … of his leadership in the development of CISE’s scientific stance on the role of computing research in energy and the environment” and “his contributions innovation in merit review.” He is also currently co-chairing visioning activities on online education for the Computing Community Consortium of the CRA.
Time: October 5, 2012
Speaker: Dave Joffe, CFA, Quantitative Research Executive
Title: Commercial Bank Assets: A Systems View
Abstract:
Finding the patterns in the network of financial relationships among bank customers
is an essential tool in sensing infrequent, cyclical and network effects in economic neighborhoods.
Insights into positive and negative contagion
in financial relationships should be added to existing risk management tools for banks
Bio:
Dave Joffe is the Quantitative Research Executive for the Information Exchange Team.
In this role, he is responsible for working with business partners to find revenue and
risk mitigation opportunities through the use of proprietary information, supporting the community
of quantitative analysts at Bank of America.
Dave and his wife Terri are parents to three children and have offered transitional care for
infants supporting adoptive families.
Dave graduated from Duke University with a B.S. in Psychology. He received his M.S. in Computer Science
from University of North Carolina Charlotte. Dave has the Chartered Financial Analyst designation.
Time: September 28, 2012
Speaker: Dr. Dangxiao Wang
Title: Towards a tangible virtual world: Haptics HCI and a surgical simulation example
Abstract:
In Human-Computer-Interaction (HCI) area, audio and visual feedback have been widely used.
However, there are still lots of open problems for haptics feedback both in fundamental
and applied research aspects. In this talk, I will explain what haptic HCI is and the motivation
to study haptic HCI. Then, framework of a haptic HCI system and key research topics will be introduced.
After a short introduction of the research work in Haptics Lab in Beihang University,
a dental simulation system for training subjects' motor skill will be elaborated.
Finally, some future research directions will be discussed.
Bio: Dangxiao WANG received the Ph.D. degree in robotics from Beihang University, Beijing, China in 2004. Currently he is an Associate Professor at the State Key Laboratory of Virtual Reality Technology and System and the Robotics Institute in Beihang University. From 2004 to 2006, he had been a post Doc at the Beihang University. After that, he was an Assistant Professor in School of Mechanical Engineering and Automation, Beihang University from 2006 to 2007. From Jan. 2012, he has been working as a visiting professor in Univerisity of North Carolina at Charlotte.
His research interests include haptic rendering, medical robotic system and haptic-based biometrics. He is a member of IEEE. He has been a Member of Executive Committee of IEEE Technical Committee on Haptics (IEEE TCH), and has served as the Vice Chair for Publications from 2011. He is the Publicity Co-Chair and Associate Editor of IEEE World Haptics Conference 2013. He served as the Award Committee member IEEE World Haptics Conference 2011. His paper on 6-DOF haptic rendering won the Best Manipulation Paper Finalist in IEEE International Conference on Robotics and Automation (ICRA 2011).
Time: September 21, 2012
Speaker: Dr. Yvo Desmedt
Title: Secure Multiparty Computation for Cloud Security
Abstract:
Although cloud computing is very popular, in February 2011, the Guardian wrote
"The speed with which Amazon and PayPal dropped WikiLeaks should be a wake-up
call to anyone who thinks that Cloud Computing services can be trusted
...". Moreover, IT Business wrote: "The countrywide Internet blackout Egypt is
experiencing may resonate with a lot of ... businesses especially as more and
more companies adopt cloud-based applications ..." Moreover, Google's policy,
i.e., they can read all your data, illustrates privacy concerns.
Although one could wonder whether CEO's have their head in the clouds instead
of both feet on the ground when they are rushing to adopt cloud technology, it
offers savings many organizations do not want to ignore.
In this talk we explain what techniques can be used to achieve reliable and
secure cloud computing and how these may bring solutions against hacking. In
particular, we introduce secret sharing and explain (1) how to use it for an
alternative to PGP for reliable and private e-mail and (2) how it plays a key
role in secure multiparty computation. We also compare homomorphic encryption
with secure multiparty computation. Moreover, we briefly discuss privacy
weaknesses in Facebook and Google+ and briefly discuss how to model privacy
using economic game theory.
Bio: Dr. Yvo Desmedt is the Jonsson Distinguished Professor at the University of Texas
at Dallas, a courtesy chair at the University College London and a Fellow of
the International Association of Cryptologic Research (IACR). He is an
(associate) editor of The Journal of Computer Security, and of Advanced
Mathematics of Communications. He is also the Editor-in-Chief of IET
Information Security and Chair of the Steering Committees of CANS and
ICITS. He was Program Chair of Crypto 1994, the ACM Workshop on Scientific
Aspects of Cyber Terrorism 2002, PKC 2003, ICITS 2007, and co-Program Chair of
CANS 2005. He was an invited speaker at conferences and workshop in 5
continents. He has authored over 200 refereed papers. He has worked on
cryptography, computer security, identification (entity authentication),
information hiding, malware, network security, and other topics.
Time: September 14, 2012
Speaker: Dr. Mary Lou Maher
Title: Designing for Collective Intelligence
Abstract:
We are facing design challenges on a much larger scale as we become an
increasingly global and technologic society. Our design solutions not only
need to respond to the needs and desires that may be included in a
specific design brief, but they also need to be environmentally
sustainable, attractive to multiple cultural groups, adaptable as
technology changes, and intuitive to potential users. Opening the design
process through crowdsourcing enables a kind of collective intelligence
that takes advantage of the diversity of the crowd, while mediating the
difficulties of collaboration. This paper presents a design space for
understanding how to design for collective intelligence that has three
dimensions: Representation, Communication, and Motivation. An analysis of
existing online innovation communities provides insight for designing for
collective intelligence. The Interactive Sustainabilty and NatureNet
projects are presented as experimental platforms for studying the design
principles for collective intelligence in public places.
Time: September 7, 2012
Speaker: Dr. John S Gero
Title: SITUATED COMPUTATION An Approach to Computational Social Science
Abstract:
Situated cognition, an emerging area of cognitive science, holds that, apart from genetics,
cognition is founded on an organism's interactions with its environment, its ability to construct memories,
and the structuring of its past experiences into situations that give meaning to what is currently experienced
and build expectations of what is being sensed. This seminar develops a set of principles for computation based
on situated cognition. It then demonstrates the application of those principles through computational
social science implementations, using cellular automata and multi-agent systems, including studying:
- social influence on behavior;
- team behavior;
- creativity and emergence;
- how individual's values change through interactions with others.
Bio: John Gero is a Research Professor at UNCC with appointments in Computer Science and Architecture,
and currently at the Krasnow Institute for Advanced Study, George Mason University. He is the author or editor
of 50 books and over 600 papers and book chapters in the fields of design science, design computing, artificial
intelligence, computer-aided design, design cognition and cognitive science. He has been a Visiting Professor of
Computer Science, Cognitive Science, Architecture, Civil Engineering, Design and Computation and Mechanical
Engineering at MIT, UC-Berkeley, UCLA,
Columbia and CMU in the USA, at Strathclyde and Loughborough in the UK, at INSA-Lyon and Provence in France and
at EPFL-Lausanne in Switzerland.
Time: August 31, 2012
Speaker: Dr. Yaorong Ge
Title: Some Projects in Health Informatics
Abstract:
Health informatics refers to information science and technologies applied
to medicine, nursing, and other healthcare fields. In contrast,
bioinformatics focuses on biology and public health informatics address
population health issues. In this talk, I will present a number of
projects that I have been working that are in the general area of health
informatics. The first project, funded by NIH/National Cancer Institute
(NCI), aims to improve the quality and efficiency of intensity modulated
radiation treatment planning using modeling and learning of past planning
data and experience. The second project, funded by NIH/National Heart,
Lung, and Blood Institute (NHLBI), is a part of a multi-institional
project that seeks to develop informatics tools and infrastructure for
longitudinal cardiovascular research. The third project, also funded by
NHLBI, focuses on innovative imaging methods for assessing cardiac risk
using non-contrast cardiac CT images. The fourth project, funded by Wake
Forest University Health Sciences, aims to develop a comprehensive
clinical data warehouse and analytical methods for clinical care
improvement and advanced decision support. I will also mention a number of
other projects that I am involved with but to a lesser degree. These
include quantitative imaging methods for cardiac injury assessment,
cardiotoxicity management in cancer survivors, and intensive care unit
(ICU) data warehouse and decision support.
