If you would like to receive announcements of these
talks, please subscribe to the
cs-seminars mailing list or subscribe to the CS Seminars RSS feed ![[RSS feed]](../Common/feed-icon16x16.png)
.
Mobility, Data Mining and Privacy: Mining Human Movement Patterns from Trajectory Data 23 Nov 2009
Speakers:Fosca Giannotti
Senior Researcher
Institute of Science and Technology Information (ISTI)
Italian National Research Council (CNR)
Knowledge Discovery and Delivery Laboratory (KDD)
Dino Pedreschi
Professor of Computer Science
Dipartimento di Informatica, University of Pisa
ISTI-CNR Knowledge Discovery and Delivery Laboratory (KDD)
Time: Monday, November 23, 2 PM Location: Babbio 221
Host: Wendy Hui Wang
ABSTRACT
The technologies of mobile communications and ubiquitous computing pervade our society, and wireless networks sense the movement of people and vehicles, generating large volumes of mobility data, such as mobile phone call records and GPS tracks. This is a scenario of
great opportunities and risks: on one side, mining this data can produce useful knowledge, supporting sustainable mobility and intelligent transportation systems; on the other side, individual privacy is at risk, as the mobility data contain sensitive personal information. A new multidisciplinary research area is emerging at this crossroads of mobility, data mining, and privacy. The talk assesses this research frontier from a data mining perspective, and illustrates the results of a European-wide research project called GeoPKDD, Geographic Privacy-Aware Knowledge Discovery and Delivery. GeoPKDD has created an integrated platform for complex analysis of mobility data, which combines spatio-temporal querying capabilities with data mining, visual analytics and semantic technologies, thus providing a full
support for the Mobility Knowledge Discovery process. In this talk, we focus on the key data mining models: trajectory patterns and trajectory clustering, and illustrate the analytical power of our system in unvealing the complexity of urban mobility in a large metropolitan area by means of a large scale experiment, based on a massive real life GPS dataset, obtained from 17,000 vehicles with on-board GPS receivers, tracked during one week of ordinary mobile activity in the urban area of the city of Milan, Italy.
BIOS
FOSCA GIANNOTTI is a senior researcher at the Information Science and Technology Institute of the National Research Council at Pisa, Italy. Her current research interests include data mining query languages, knowledge discovery support environment, web-mining, spatio-temporal reasoning, spatio- temporal data mining, and privacy preserving data mining. She has been the coordinator of various European and national research projects and is currently the co-ordinator of the FP6- IST project GeoPKDD: Geographic Privacy-aware Knowledge Discovery and Delivery. She is responsible for the Working Group on Privacy and Security in Data mining of the KDUBIQ network of excellences. She has taught classes on databases and data mining at universities in Italy and abroad. She is the author of more than one hundred publications and served in the scientific committee of various conferences in the area of Logic Programming, Databases, and Data Mining. In 2004 she co-chaired the European conference on Machine Learning and Knowledge
Discovery in Data Bases ECML/PKDD 2004. She is the program co-chair of ICDM 2008, the IEEE Int. Conf. on Data Mining.
Publications of Fosca Giannotti from DBLP bibliographic server:
http://www.informatik.uni-trier.de/~ley/db/indices/a-tree/g/Giannotti:Fosca.html
DINO PEDRESCHI is a full professor at the Dipartimento di Informatica of the University of Pisa. He has been a visiting scientist and professor at the University of Texas at Austin (1989/90), at CWI Amsterdam (1993) and at UCLA (1995). His current research interests are in data mining and logic in databases, and particularly in data analysis, in spatio-temporal data mining, and in privacy-preserving data mining. He has taught classes on programming languages, databases and data mining in universities in Italy and abroad. He is a
member of the program committee of the main international conferences on data mining and knowledge discovery.
He has been a co-chair of ECML/PKDD 2004, the international conference on Machine Learning and Knowledge Discovery in Databases. He served as the coordinator of the undergraduate studies in Computer Science at the University of Pisa, and as a vice-rector of the same university, with responsibility in teaching affairs. He is the national coordinator of the project "GeoPKDD.it", Geographic Privacy-aware Knowledge Discovery and Delivery.
Fosca Giannotti and Dino Pedreschi co-lead the Pisa KDD Lab -Knowledge Discovery and Data Mining Laboratory (http://www-kdd.isti.cnr.it) - a joint research initiative of the University of Pisa and the Information Science and Technology Institute of the Italian National Research Council, founded in 1995, one of the earliest European research groups specifically targeted at data mining and knowledge discovery.
Publications of Dino Pedreschi from DBLP bibliographic server:
http://www.informatik.uni-trier.de/~ley/db/indices/a-tree/p/Pedreschi:Dino.html
Upcoming Seminars
Dec. 2: Bjarne Stroustrup
Dec. 7: Jan Allbeck
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Excursions in Computation 16 Nov 2009
Speaker: Wayne Patterson, Professor of Computer Science, Howard UniversityTime: Monday, November 16, 2 PM Location: Babbio 221
Host: Susanne Wetzel
Abstract:
The author is reminded of the old expression: "Something old, something new; something borrowed, something blue." Although reluctant to suggest a presentation anything like a wedding ceremony, he will look anew at some old computational concepts involving the Pascal
triangle; something new (to many) in a related application revisiting a public key crypto chestnut; borrowing some ideas from what is now usually described as "experimental mathematics". Something blue? You'll have to wait and see.
Bio:
Wayne Patterson was born in Moncton, New Brunswick, Canada. He received the B. Sc. (Honours) in Applied Mathematics at the University of Toronto in 1966; M. Sc. in Mathematics also from Toronto in 1967; and the Ph. D. in Mathematics from the University of Michigan in 1971, in the field of differential topology. He also later received the M. Sc. in Computer Science from the University of New Brunswick in 1982. He was a Post-Doctoral Fellow in 1971-2 at Princeton University and in 1972-73 at the University of California at Berkeley.
In 1968, he began to teach in an exemplary project called Project SEED, which teaches advanced mathematics (high school and college level algebra and calculus) to inner-city at-risk students throughout the United States and elsewhere. He taught in this project and became its Associate National Director until 1975; although he remained on its Board of Directors and has now served as the Chair of Project SEED’s Board of Directors for the past 24 years. The SEED Project, now in its 46th year, continues to serve tens of thousands of at-risk students annually.
In 1975, he returned to Canada to join the Government of Canada, serving as Special Assistant and Economic Advisor to the Secretary of State and later the Deputy Prime Minister of Canada.
He returned to higher education as a professor of mathematics and later computer science at the Université de Moncton, the only French-language university in Canada outside of Québec. While teaching at Moncton, he also was a candidate for the House of Commons of Canada, and the Legislature of the Province of New Brunswick; and was twice elected as the national Vice-President of the Liberal Party of Canada.
In 1984, Dr. Patterson was appointed Chair of the Department of Computer Science at the University of New Orleans, and in 1988 Associate Vice Chancellor for Research at that university. In 1993, he was appointed Vice President for Research and Professional and Community Services, and Dean of the Graduate School at the College of Charleston and the University of Charleston, South Carolina.
In 1998, he was selected by the Council of Graduate Schools, the national organization of graduate deans and graduate schools, as the Dean in Residence at the national office in Washington, DC. His other service to the graduate community in the United States has included being elected to the Presidency of the Conference of Southern Graduate Schools, and also to the Board of Directors of the Council of Graduate Schools.
Since the year 2000, he has been the Senior Fellow for International Programs and Academic Program Review in the Graduate School at Howard University, and Professor of Computer Science at Howard as well. From December 2003 to July 2006, he also served as Associate Vice Provost for Research at Howard.
In his own disciplinary research, Dr. Patterson has published more than 50 scholarly articles, and a leading textbook, Mathematical Cryptology (Rowman and Littlefield, 1986). He has been the principal investigator on over 35 external grants valued at over $6,000,000.
In August 2006, he was loaned by Howard University to the US National Science Foundation to serve as the Foundation’s Program Manager for International Science and Engineering in Developing Countries. He completed this three-year rotation in August 2009 after having negotiated an agreement for joint funding of research in developing countries with USAID, and having made in 2009 the largest volume ever of NSF research projects involving Africa. He has now returned to the Graduate School staff at Howard.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Using Cloud Computing Technology for Defensive Data Analysis 2 Nov 2009
Speaker: Neal Ziring, Technical Director for Vulnerability Analysis & Operations, NSATime: Monday, November 2, 2PM
Location: Babbio 221
Host: Susanne Wetzel
Abstract:
US computer networks are under attack by increasing sophisticated adversaries. Network defenders, such as the NSA Blue Team, have extensive sources of information about hosts and networks available. But the volume of the data and its complexity make analysis with
conventional methods very slow and inflexible. This talk will present results from applying open-source cloud computer technologies, including the Map-Reduce parallel computing model, to large-scale network defense data analysis. These technologies, already in use by
many Internet companies, offer the flexibility and scale to improve network defense capabilities and reduce response times.
Bio:
Neal Ziring is a Defense Intelligence Senior Level computer scientist with the NSA. He joined NSA in 1989, and has spent his time there mostly in working in security evaluations and security guidance. Since 1996, he has worked in network and protocol security, and in security architecture for NSA mission systems. He currently serves as a technical director in the NSA Information Assurance Directorate. Prior to joining NSA, Neal worked on software tools at AT&T Bell Labs. He has an MS in Computer Science and a BS in Electrical Engineering, both from Washington University in St. Louis.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Usable Security Lessons for Creating Effective Browser Warnings 19 Oct 2009
Speaker: Serge Egelman, BrownTime: Monday, October 19, 2PM
Location: Babbio 221
Host: Sven Dietrich
Abstract :
In a world where making an incorrect online trust decision can mean the difference between checking your account balance and transferring it to criminals, Internet users need effective security warnings to help them identify risky situations. In a perfect world, software could automatically detect all security threats and then block access to high risk websites. Because there are many threats that we cannot detect with 100% accuracy and false positives are all too frequent, web browser vendors generally opt to warn users about security threats. In this talk I cover the common pitfalls of web browser security warnings and draw parallels with warnings in the physical world. I describe the results of laboratory phishing studies I performed in order to examine users' mental models, risk perceptions, and comprehension of current security warnings. Finally, I show how I used these findings to design and test a more usable SSL warning that better conveys risk and uses context to minimize habituation effects.
Bio:
Serge Egelman is a postdoctoral researcher at Brown University working on access control mechanisms that minimize human error. He also dabbles in behavioral economics in order to better understand why people make poor security choices. He recently earned a PhD from Carnegie Mellon University's School of Computer Science. His main research area is on usable privacy and security, which has included work on phishing detection, authentication systems, online privacy, user account models, and online shopping behaviors. Serge was a summer intern at PARC in 2006, as well as an intern at Microsoft Research for six months in 2008. While at MSR, he helped the IE team redesign the IE8 phishing warning based on the results of his dissertation research. Serge enjoys traveling the world and hopes to visit every UNESCO World Heritage Site.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Does Privacy Require True Randomness? 5 Oct 2009
Speaker: Carl Bosley (Stevens)Time: Monday, October 5, 2PM
Location: Babbio 221
Host: Antonio Nicolosi
Abstract:
Most cryptographic primitives require randomness (for example, to generate secret keys). Usually, one assumes that perfect randomness is available, but, conceivably, such primitives might be built underweaker, more realistic assumptions. This is known to be achievable for many authentication applications, when entropy alone is typically sufficient. In contrast, all known techniques for achieving privacy seem to fundamentally require (nearly) perfect randomness. We ask the question whether this is just a coincidence, or, perhaps, privacy inherently requires true randomness? We completely resolve this question for information-theoretic private-key encryption, where parties wish to encrypt a b-bit value using a shared secret key sampled from some imperfect source of randomness S . Our technique also extends to related primitives which are sufficiently binding and hiding, including computationally secure commitments and public-key encryption. Our main result shows that if such n-bit source S allows for a secure encryption of b bits, where b > log n, then one can deterministically extract nearly b almost perfect random bits from S. Further, the restriction that b > log n is nearly tight: there exist sources S allowing one to perfectly encrypt (log n - loglog n) bits, but not to deterministically extract even a single slightly unbiased bit. Hence, to a large extent, true randomness is inherent for encryption: either the key length must be exponential in the message length b, or one can deterministically extract nearly b almost unbiased random bits from the key. In particular, the one- time pad scheme is essentially "universal".
Bio:
Carl Bosley received an A.B. from Harvard in 2001 in Mathematics, and a Ph.D. from New York University in 2009 in Computer Science, under the supervision of Yevgeniy Dodis. His interests include randomness requirements for cryptographic applications, and relationships between hardness of learning and cryptography.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Automated Detection of Stealth Attacks on the Operating System Kernel 28 Sep 2009
Speaker: Arati Baliga, Rutgers
Time: Monday, September 28, 2PM
Location: Babbio 221
Host: Vivek Pathak
Abstract:
The operating system kernel is implicitly trusted by applications running on a computer system. An attack on the operating system kernel that alters its state is critical because it puts all applications at risk. A compromised system can be stealthily exploited by the attackers, in several ways, such as exfiltration of sensitive information, wasteful usage of the system's resources, adversely affecting system performance or involving it in fraudulent or malicious activities without the user's knowledge or permission. The lack of appropriate detection tools allows such systems to lie within the attackers' control for indefinite periods of time.
Stealth attacks on the kernel are carried out by malware commonly known as rootkits. Though rootkits have considerably increased in sophistication over the past few years, their primary purpose is to conceal the presence of the attacker and therefore, focus on hiding user level objects. In this talk, I will present a new class of stealth attacks on the kernel that we have identified, which do not attempt to hide objects but are inherently stealthy by design. They achieve their malicious objectives by solely modifying data within the kernel. I will also describe an automated technique that can be used for detection of such stealthy data-centric attacks. The key idea behind this technique is to automatically identify and extract invariants exhibited by kernel data structures during a training phase on a clean kernel. The hypothesis is that rootkits that manipulate kernel data violate some of
these invariants and therefore, can be detected. These inferred invariants are then used as specifications of data structure integrity and are enforced during runtime.
Bio:
Arati Baliga is a Research Associate at the Wireless Information Network Laboratory (WINLAB), Rutgers University. Her current research includes improving the security and reliability of application programs using transactional memory and securing cognitive radio networks. She completed her Ph.D in January 2009 from the department of Computer Science at Rutgers. Her research interests span system security, security in wireless and emerging networks, operating systems and distributed systems.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Efficient Training of Classifiers for Detection 14 Sep 2009
Speaker: Alex Berg, Columbia University
Time: Monday, September 14, 2PM
Location: Babbio 221
Host: Philippos Mordohai
Abstract:
Computer vision researchers are presented with a potentially embarrassing problem these days -- too much data. In particular very large data sets can be a challenge for learning algorithms used in computer vision. We have shown that a class of techniques used in computer vision that had previously been considered prohibitively expensive can in fact be used very efficiently. This result comes in two parts, first determining the form of the classifiers in question and demonstrating efficient evaluation, and second showing how to train such classifiers efficiently.
Part 1
Straightforward classification using kernelized SVMs requires evaluating the kernel for a test vector and each of the support vectors. For additive kernels we show that one can do this much more efficiently. In particular we show that one can build histogram intersection kernel SVMs (IKSVMs) with runtime complexity of the classifier logarithmic in the number of support vectors as opposed to linear for the standard approach. We further show that by precomputing auxiliary tables we can construct an approximate classifier with constant runtime and space requirements, independent of the number of support vectors, with negligible loss in classification accuracy on various tasks. This approximation also applies to 1-Chi squared and other kernels of additive form.
Classification using Intersection Kernel Support Vector Machines is Efficient
S. Maji, A.C. Berg, J. Malik
CVPR 2008
Part 2
We present methods for training high quality object detectors very quickly. The core contribution is a pair of fast training algorithms for piece-wise linear classifiers, which can approximate arbitrary additive models. The classifiers are trained in a max-margin framework and significantly outperform linear classifiers on a variety of vision data sets. We report experimental results quantifying training time and accuracy on image classification tasks and pedestrian detection, including detection results better than the best previous on the INRIA pedestrian dataset with faster training.
Max-Margin Additive Classiffiers for Detection
S. Maji & A.C. Berg
ICCV 2009
Bio:
Alex Berg's research concerns computational visual recognition. He has worked on general object recognition in images, action recognition in video, human pose identification in images, image parsing, face recognition, image search, and machine learning for computer vision.
He is currently a research scientist at Columbia University. Prior to that he was a research scientist in Yahoo! Research and a visiting scholar at U.C. Berkeley. His PhD at U.C. Berkeley developed a novel approach to deformable template matching. He earned a BA and MA in Mathematics from Johns Hopkins University and learned to race sailboats at SSA in Annapolis.
Time: Monday, September 14, 2PM
Location: Babbio 221
Host: Philippos Mordohai
Abstract:
Computer vision researchers are presented with a potentially embarrassing problem these days -- too much data. In particular very large data sets can be a challenge for learning algorithms used in computer vision. We have shown that a class of techniques used in computer vision that had previously been considered prohibitively expensive can in fact be used very efficiently. This result comes in two parts, first determining the form of the classifiers in question and demonstrating efficient evaluation, and second showing how to train such classifiers efficiently.
Part 1
Straightforward classification using kernelized SVMs requires evaluating the kernel for a test vector and each of the support vectors. For additive kernels we show that one can do this much more efficiently. In particular we show that one can build histogram intersection kernel SVMs (IKSVMs) with runtime complexity of the classifier logarithmic in the number of support vectors as opposed to linear for the standard approach. We further show that by precomputing auxiliary tables we can construct an approximate classifier with constant runtime and space requirements, independent of the number of support vectors, with negligible loss in classification accuracy on various tasks. This approximation also applies to 1-Chi squared and other kernels of additive form.
Classification using Intersection Kernel Support Vector Machines is Efficient
S. Maji, A.C. Berg, J. Malik
CVPR 2008
Part 2
We present methods for training high quality object detectors very quickly. The core contribution is a pair of fast training algorithms for piece-wise linear classifiers, which can approximate arbitrary additive models. The classifiers are trained in a max-margin framework and significantly outperform linear classifiers on a variety of vision data sets. We report experimental results quantifying training time and accuracy on image classification tasks and pedestrian detection, including detection results better than the best previous on the INRIA pedestrian dataset with faster training.
Max-Margin Additive Classiffiers for Detection
S. Maji & A.C. Berg
ICCV 2009
Bio:
Alex Berg's research concerns computational visual recognition. He has worked on general object recognition in images, action recognition in video, human pose identification in images, image parsing, face recognition, image search, and machine learning for computer vision.
He is currently a research scientist at Columbia University. Prior to that he was a research scientist in Yahoo! Research and a visiting scholar at U.C. Berkeley. His PhD at U.C. Berkeley developed a novel approach to deformable template matching. He earned a BA and MA in Mathematics from Johns Hopkins University and learned to race sailboats at SSA in Annapolis.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
From Dynamic to Static and Back: Riding the Roller Coaster of Information-Flow Control Research 6 Jul 2009
Speaker: Andrei Sabelfeld, Chalmers University of TechnologyTime: Monday, July 6, 3PM
Location: Babbio 304
Host: David Naumann
Abstract:
Historically, dynamic techniques are the pioneers of the area of information flow in the 70’s. In their seminal work, Denning and Denning suggest a static alternative for information-flow analysis. Following this work, the 90’s see the domination of static techniques for information flow. The common wisdom appears to be that dynamic approaches are not a good match for security since monitoring a single path misses public side effects that could have happened in other paths. Dynamic techniques for information flow are on the rise again, driven by the need for permissiveness in today’s dynamic applications. But they still involve nontrivial static checks for leaks related to control flow. This talk demonstrates that it is possible for a purely dynamic enforcement to be as secure as Denning-style static information-flow analysis, despite the common wisdom. We do have the trade-off that static techniques have benefits of reducing runtime overhead, and dynamic techniques have the benefits of permissiveness (this, for example, is of particular importance in dynamic applications, where freshly generated code is evaluated). But on the security side, we show for a simple imperative language that both Denning-style analysis and dynamic enforcement have the same assurance: termination-insensitive noninterference.
Joint work with Alejandro Russo, Chalmers University of Technology
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Robust Decentralized Authentication in Peer-to-peer, Social, and Ad-hoc Networks 4 May 2009
Speaker: Vivek Pathak (Rutgers)Time: Monday, May 4, 2PM
Location: Babbio 221
Host: Philippos Mordohai
Abstract:
Authentication has traditionally been done either in a decentralized manner with human assistance or automatically through a centralized security infrastructure. In the security infrastructure approach, a central trusted authority takes on the responsibility of authenticating participants within its domain of control. While the security infrastructure approach works well in traditional organizations, it does not address the needs of open membership systems.
We propose automatic decentralized authentication mechanisms for peer-to-peer networks, social networks, and ad-hoc networks. Our byzantine fault tolerant public-key authentication protocol (BPKA) provides decentralized authentication to peer-to-peer systems with honest majority. Authentication is done over an insecure asynchronous network without using trusted third parties or human input. We also authenticate public keys in the email environment through our social-group key authentication protocol (SGKA). The protocol provides end-to-end authentication at the email client without using infrastructure or centralized authorities. Finally, location authentication in ad-hoc networks is proposed through our geographical secure path routing protocol (GSPR). The protocol authenticates geographic locations of anonymous nodes in order to provide location authentication and anonymity simultaneously.
Bio:
Vivek Pathak obtained the Ph.D. degree in Computer Science from Rutgers University in 2008. The focus of his research was on securing peer-to-peer, socially networked, and ad-hoc systems. The title of his dissertation was Robust Decentralized Authentication for Public Keys and Geographic Location. His research interests include infrastructure free security and ecommerce for decentralized systems, social networks, and the Internet.
Vivek Pathak is currently working for Ask.com (http://www.ask.com/) as the technical product manager responsible for web search. He has previously worked in software engineering roles for Ask.com and i2 technologies. He also has the B. Tech. degree in Aerospace Engineering from the Indian Institute of Technology, Bombay, and the M.B.A. degree from New York University. More information can be found at http://paul.rutgers.edu/~vpathak.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Machine Learning with Graphs and Matchings 27 Apr 2009
Speaker: Tony Jebara, Columbia UniversityTime: Monday, April 27, 2PM
Location: Babbio 221
Host: Philippos Mordohai
Abstract:
Many machine learning problems on data can naturally be formulated as problems on graphs. For example, dimensionality reduction and visualization are related to graph embedding. Given a sparse graph between N high-dimensional data nodes, how do we faithfully embed it in low dimension? We present an algorithm that improves dimensionality reduction by extending semidefinite embedding methods. But, given only a dataset of N samples, how do we construct a sparse graph in the first place? The space to explore is daunting with 2^(N2) graphs to choose from yet two interesting subfamilies are tractable: matchings and b-matchings. By placing distributions over matchings and using loopy belief propagation, we efficiently infer the optimal graph. Matching not only has intriguing algebraic properties, it also leads to improvements in graph reconstruction, graph embedding, graph labeling, and graph partitioning. We show results on text, network and image data. Time permitting, we will show results on location data from millions of tracked mobile phone users which lets us discover patterns of human behavior, networks of places and networks of people.
Bio:
Tony Jebara is Associate Professor of Computer Science at Columbia University and director of the Columbia Machine Learning Laboratory. His research intersects computer science and statistics to develop new frameworks for learning from data with applications in vision, networks, spatio-temporal data, and text. Tony is also co-founder of Sense Networks. He has published over 50 peer-reviewed papers in conferences and journals including NIPS, ICML, UAI, COLT, JMLR, CVPR, ICCV, and AISTAT. He is the author of the book Machine Learning: Discriminative and Generative. Tony is the recipient of the Career award from the National Science Foundation and has also received honors for his papers from the International Conference on Machine Learning and the Pattern Recognition Society. Tony's research has been featured on television (ABC, BBC, New York One, TechTV, etc.) as well as in the popular press (New York Times, Slash Dot, Wired, Scientific American, Newsweek, etc.). He obtained his PhD in 2002 from MIT. Recently, Esquire magazine named him one of their Best and Brightest of 2008. Tony's lab is supported in part by the NSF, CIA, NSA, DHS, and ONR .
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Making Fake Skin (and other things) Look Real 13 Apr 2009
Speaker: Craig Donner, Columbia UniversityTime: Monday, April 13, 2PM
Location: Babbio 221
Host: Philippos Mordohai
Abstract:
Natural materials such as juice, leaves, marble, and skin have a complex interaction with light. Light refracts into such materials and scatters many times before exiting, and this subsurface scatteringof light has a profound impact on their appearance. In this talk, I will present recent advances in computer graphics for making images and acquiring the appearance of natural materials, using examples including liquids, leaves, and especially human skin.
Bio:
Craig Donner is currently a Postdoctoral Research Scientist at Columbia University. His core research interests are appearance modeling and global illumination in the context of photorealistic image synthesis, as well as the acquisition and modeling of light transport in complex materials. His work is now used in the film, game, and even cosmetics industries. Craig received his doctorate at the University of California, San Diego, investigating the scattering of light in translucent materials.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Distributed Control of Networked Robots and Systems 6 Apr 2009
Speaker: Michael Zavlanos, University of PennsylvaniaTime: Monday, April 6, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Philippos Mordohai
Abstract:
The field of robotics is evolving from single monolithic robots to teams of small but interconnected robots achieving global objectives using local coordination. Coordinated missions for teams of mobile robots include coordinated estimation, surveillance, and coverage, coordinated satellite alignment and synchronization, as well as distributed placement and assignment in creating desirable team structures. The fundamental challenge in such problems is the design of local rules, such as distributed controllers and estimators, which by local coordination give rise to the desired global objectives.
In this talk, I will first present the first distributed, scalable, and verifiable algorithm that allows teams of robots to dynamically create any desired structure, characterized by the relative locations of the robots in it, using local coordination rules. This is achieved using a combination of multi-destination potential fields and assignment coordination protocols. I will then address the problem of maintaining connectivity in robotic networks, where the robot nodes are mobile. The proposed solution is not only the first distributed solution to this problem, but has been both theoretically and experimentally verified, and can be composed with other objectives to give robotic structures that can adapt to environmental changes and handle node failures.
Bio:
Michael M. Zavlanos received the Diploma in mechanical engineering from the National Technical University of Athens, Athens, Greece, in 2002 and the M.S.E. and Ph.D. degrees in electrical and systems engineering from the University of Pennsylvania, Philadelphia, in 2005 and 2008, respectively. He is currently a Postdoctoral Researcher with the Department of Electrical and Systems Engineering, University of Pennsylvania. His current research interests include the areas of distributed control systems, networked systems, and hybrid dynamical systems with applications to robotics, sensor networks, and biomolecular networks. Dr. Zavlanos
was a finalist of the Best Student Paper Award at the 45th IEEE Conference on Decision and Control in 2006.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Civitas: Toward a Secure Voting System 30 Mar 2009
Speaker: Michael Clarkson, CornellTime: Monday, March 30, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: David Naumann
Abstract:
Voting systems are hard to make trustworthy because they have strong, conflicting security requirements: Voters must be convinced that their votes are tallied correctly, while the secrecy of those votes must also be maintained---even when someone tries to buy votes or physically coerce voters. This talk presents Civitas, an electronic remote voting system satisfying these requirements and offering assurance through both cryptographic security proofs and information-flow analysis.
Bio:
Michael Clarkson is a PhD candidate in the Department of Computer Science at Cornell University. For more information, please visit http://www.cs.cornell.edu/people/clarkson/
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Detection, Tracking and Registration in 3D Radiology Image Analysis 23 Mar 2009
Speaker: Lin Yang, RutgersTime: Monday, March 23, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Philippos Mordohai
Abstract:
In this talk, we will describe a robust, fast and accurate 3D object detection/tracking algorithm which is developed for the 3D echocardiography. According to our knowledge, this is the first study reporting fast and reliable 3D ultrasound tracking of the left ventricle on a very large dataset, which contains 1143 3D volumetric data. From our research we report that collaborative trackers increased the tracking accuracy dramatically. The final accurate results are achieved by applying the motion priors using one-step forward prediction on the manifold. The robustness to complex background and weak edges come from the learned discriminative detectors and boundary classifiers, while the temporal consistence is preserved by template tracker. Instead of building specific models for heart, all the major steps in our algorithm are based on learning. Our proposed algorithm is therefore general enough to be extended to other 2D/3D medical image object detection/tracking problems.
We will also present a new method for fast and robust image registration combining landmark and region based techniques. The algorithm is completely unsupervised and computationally efficient. Due to a relatively small number of landmarks, the method runs faster than many existing nonlinear registration algorithms reported in the literature, such as B-Spline based image registration and the Demon's algorithm. The method can also handle large transformation and deformation while still providing good registration results. We will also explain how to implement the algorithm in a multi-core platform, the IBM Cell Broadband Engine.
Bio:
Lin Yang received his Ph. D. degree in the department of Electrical and Computer Engineering, Rutgers University in 2009.
He will serve as an assistant professor in the department of Radiology in the University of Medicine and Dentistry of New Jersey and also hold a joint assistant professor appointment in the department of Biomedical Engineering in Rutgers in August, 2009. His research interests include different areas of medical image analysis, computer vision and machine learning. He is working on the design and development of content-based image and video retrieval and 2D/3D medical image analysis including detection, segmentation, registration and tracking. He has already published over 20 peer-reviewed articles. For detailed information, please visit http://www.eden.rutgers.edu/~linyang/.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Efficient Comparison of 3D Models for Shape Retrieval 16 Mar 2009
Speaker: Ameesh Makadia, Google ResearchTime: Monday, March 16, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Philippos Mordohai
Abstract:
The ability to perform fast and accurate retrieval from a database of 3D models is becoming a growing necessity as the number of models in circulation is rapidly increasing. Most search engines, built around text-based search, fail to leverage shape content, which often leads to search results of limited success and applicability. In this work we explore the problem of retrieval from large databases of 3D models
that are queried by example models.
At the core of this retrieval task is the fundamental challenge of defining and evaluating similarity between 3D shapes. I will present a method for comparing models that is based on a visual representation. Specifically, we will consider collections of rendered images from numerous viewpoints surrounding the model. While rendered images provide sufficiently discriminating information, similarity computation requires consideration of all possible rotational alignments between models. The main novelty is in the realization that the comparison of visual representations can be expressed as a correlation of spherical functions, and we will show how evaluation can be performed efficiently using techniques from spherical signal processing. Qualitative and quantitative retrieval results will be shown on a large web dataset of over 1 million 3D models.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Property Verification of an Electronic Payment System: EP2 2 Mar 2009
Speaker: Temesghen Kahsai, Dep. of Computer Science, Swansea University, UKTime: Monday, March 2, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: David Naumann
Abstract:
The EP2 system is an electronic payment system and it stands for 'EFT/POS 2000' short for 'Electronic Fund Transfer / Point of Service 2000', is a joint project established by a number of (mainly Swiss) financial institutes in order to define the infrastructure for credit, debit and electronic purse terminals in Switzerland (www.eftpos2000.ch). The system consists of seven autonomous entities and they are centered around an EP2 Terminal. These entities communicate with the Terminal and, to a certain extent, with one another via XML-messages in a fixed format. Each component is a reactive system defined by a number of use cases. The EP2 Specification consists of 12 documents, each of which describe the different components or some aspect common to the components.
In this talk I will show the modeling of the EP2 specification in the formal specification language CSP-CASL [2]. CSP-CASL allows to formalize computational system in a combined algebraic / process algebraic notation. In [1] we have developed a proof method for various refinement notions of CSP-CASL. Using such proof method we verify the refinement of the different level of the EP2 specification. We also verify properties such as deadlock and livelock freedom using an interactive theorem prover.
References:
[1] T. Kahsai and M. Roggenbach. "Property preserving refinement for CSP-CASL". In A.Corradini and U. Montanari. Recent Trends in Algebraic Development Techniques. Springer-Verlang. To appear.
[2] M. Roggenbach. "CSP-CASL -- A new integration of process algebra and algebraic specification" Theoretical Computer Science, 354:42-71. 2006
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Controlling Sensors through Physics: some Ideas for the well-founded Control of Mobile Sensor Networks 23 Feb 2009
Speaker: Simon Dobson, UCD Dublin IETime: Monday, Feb 23, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Dominic Duggan
Abstract:
Mobile sensors are an attractive proposition for environmental sensing, but pose significant engineering problems. Not least amongst these is the need to match the behaviour of the sensor platform to the physical environment in which it operates. We present initial work on using models of physical processes to generate models for autonomic control, and speculate that these can be used to improve the confidence we can place in sensed data.
Bio:
Simon Dobson has a research career spanning over fifteen years in academia, government and industry. His research centres around adaptive pervasive computing and novel programming techniques, addressing both theory and practice and being supported by an extensive record of published work (including papers in CACM, TAAS, JPDC, EHCI and ECOOP) and primary authorship on grants worth over EUR 3M (and further involvements grants worth over EUR 28M) feeding around EUR 1.5M directly into his own current research programme. His expertise is widely recognised internationally: he serves on the steering or programme committees of many international conferences and workshops including PMCJ, PERVASIVE, AN, ICAC, ICOST, ECOOP, SAPIR, MUCS and MPAC; is a reviewer for journals including ACM Transactions on Autonomous and Adaptive Systems, SOFTWARE - Practice and Experience, and IEEE Communications; has been an invited editor for special issues of Computer Networks, IJIPT and JNSM; and participates in a number of EU strategic workshops and working groups. He is National Director for the European Research Consortium for Informatics and Mathematics, a board member of the Autonomic Communication Forum (at which he chairs the semantics working group), and a member of the IBEC/ICT Ireland standing committee on academic/industrial research and development. As a co-founder and CEO of a research-led start-up company he has experience in steering basic research to commercialisation. He holds a BSc and DPhil in computer science, is a Chartered Engineer and Chartered IT Professional, and member of the BCS, IEEE and ACM.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Efficient Image Search and Retrieval using Compact Binary Codes 9 Feb 2009
Speaker: Rob Fergus, NYUTime: Monday, Feb 9, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Philippos Mordohai
Abstract:
The vast majority of information on the Internet is in visual form, yet we currently lack effective methods for searching images or videos. Existing strategies rely mainly on textual cues which give impoverished and often misleading descriptions of the visual content. A key part of the challenge is that the search needs to be highly efficient due to the scale of the problem: Google's Image Search indexes around 10 billion images, while YouTube holds petabytes of video data and receives 10 hours of new content each minute.
In my talk I will describe methods for efficiently searching Internet-sized image databases. Using machine learning techniques, we represent each image with a compact binary code, at most a few hundred *bits* in length, which preserves the original neighborhood structure of images in the database. Our scheme is able to perform real-time search on millions of images using a standard PC, obtaining a retrieval performance comparable with that of more complex descriptors, despite being many orders of magnitude faster.
Joint work with: Antonio Torralba (MIT), Yair Weiss (Hebrew University).
Bio:
Rob Fergus is currently an Assistant Professor of Computer Science at the Courant Institute of Mathematical Sciences, New York University. Originally from the UK, Fergus did his undergraduate degree in Electrical Engineering at the University of Cambridge. He then did a Masters in Electrical Engineering with Prof. Pietro Perona at Caltech, before completing a PhD with Prof. Andrew Zisserman at the University of Oxford. Before coming to NYU, Fergus spent two years as a post-doc in the Computer Science and Artificial Intelligence Lab (CSAIL) at MIT, working with Prof. William Freeman. In 2003, he and his co-authors were awarded the CVPR Best Paper Prize. In 2005, his PhD thesis won the prize for the best Computer Science thesis in the UK.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Developing Secure Software - Builders vs. Breakers 2 Feb 2009
Speaker: Boaz Gelbord, Wireless GenerationTime: Monday, Feb 2, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Wendy Hui Wang
Abstract:
What is the best way to produce secure code - through Building or Breaking? The Building school holds that secure software can only result from security-trained developers who incorporate security principles into their coding. On the other hand, many software shops are built around a Breaking model - build software as fast as possible and then try to have experts break it at some point in the software development lifecycle.
In this talk we examine this issue in the wider context of enterprise security risk management and by looking at particular web application vulnerabilities. Topics we will cover include: - An overview of enterprise security risks
- Compliance driven security measures
- The commoditization of network security and increased importance of application security risk
- Resource allocation
- Specific challenges of securing web applications
- Analysis of vulnerability lists like the OWASP Top 10 and CWE/SANS Top 25
Bio:
Dr. Boaz Gelbord is the Executive Director of Information Security at Wireless Generation, a New York based educational technology company and a leading provider of assessment tools in thousands of classrooms across the country.
Boaz began his career as a cryptologist at KPN Royal Dutch Telecom, where he led numerous security projects and authored 12 patents relating to information security. His work on privacy enhancing technologies at KPN earned several international awards and led to his designation as one of "Europe's Tech Stars" by the Wall Street Journal Europe. Boaz was appointed as an independent expert to the eEurope 2005 Advisory Group, a high level committee that advised the European Commission on Internet policy. He was also an appointed expert in several ETSI (European Telecommunications Standards Institute) Specialist Task Forces, including the Secure Algorithm Group of Experts that standardized the GSM and UMTS encryption algorithms. Boaz taught information security for several years as an Associate Professor at the University of Leiden in the Netherlands.
Boaz was one of the founders of the European Network and Information Security Agency, the official EU body responsible for information security where he headed the Security Technologies Unit. Boaz has
chaired the program and steering committees of several leading international security conferences, including being the co-Chairman of the Steering Committee of the ISSE 2005 conference, Europe's largest
independent information security conference.
Boaz was also the first Director of Information Security at the New School in New York City where he introduced and implemented a comprehensive information security program. He holds a BSc in mathematics from the University of Calgary, an MSc in mathematics from the University of Toronto, and a PhD in mathematics from the Technion in Israel. He hold the CISA and CISSP certifications and is a frequently invited keynote speaker on information security and privacy issues. In the past he has presented keynote addresses at ITU, CEN, ETSI, and East West Institute conferences.
Boaz blogs on information security at www.boazgelbord.com
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Controlling Timing Channels in Multithreaded Programs 25 Nov 2008
Speaker: Alejandro RussoChalmers University of Technology, Gothenburg, Sweden
Time: Monday, Dec 1, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Dave Naumann
Abstract:
The problem of controlling information flow in multithreaded programs remains an important open challenge.A major difficulty for tracking information flow in concurrent programs is due to the internal timing covert channel. Information is leaked via this channel when secrets affect the timing behavior of a thread, which, via the scheduler, affects the interleaving of public events. This channel is particularly dangerous because, in contrast to external timing, the attacker does not need to observe the actual execution time of programs. This talk introduces a novel treatment of the interaction between threads and the scheduler. As a result, a permissive security specification and a compositional security type system are obtained. The type system guarantees security for a wide class of schedulers and provides a flexible treatment of dynamic thread creation and synchronization. The approach relies on the modification of the scheduler in the run-time environment. In some scenarios, the modification of the run-time environment might not be an acceptable requirement. For such scenarios, we briefly present two transformations that eliminate the need for modifying the scheduler while avoiding internal timing leaks. The first transformation is given for programs running under cooperative schedulers. It states that threads must not yield control inside of computations that branch on secrets. The second transformation closes internal timing channel when the scheduler is preemptive and behaves as round-robin. It spawns dedicated threads, whenever computation may affect secrets, and carefully synchronizes them.
Talk based on Alejandro Russo's PhD thesis available at http://www.cs.chalmers.se/~russo/russothesis.pdf
Talk based on Alejandro Russo's PhD thesis available at http://www.cs.chalmers.se/~russo/russothesis.pdf
Short Bio:
Alejandro Russo is a PhD Student under the supervision of Andrei Safelfeld at the Chalmers University of Technology in Sweden. Alejandro will defend his PhD thesis this year on November 13th. Prior to establishing in Chalmers, Alejandro spent a few months as a research assistant at the Stevens Institute of Technology in USA. Before that, He obtained the degree of Licentiate in Computer Science (five years degree equivalent to master) at the National University of Rosario in Argentina.
Alejandro main research area is language-based security, while programming languages and computer security are his general research interests. He has been working with preservation of confidentiality of data in concurrent languages as well as the design of libraries to guarantee security policies in functional languages.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Authenticated Communication in Dynamic Federated Systems 23 Nov 2008
Speaker: Nelly FazioDepartment of Computer Science
City University of New York
City College and Graduate Center
Time: Monday, Nov 24, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Wendy Hui Wang
Abstract:
System F6 is a DARPA program to increase the value and flexibility of future satellite missions. To this aim, the F6 initiative suggests replacing the monolithic architecture of current spacecrafts with a cluster of small specialized satellites. Realizing this vision of a collaborating federation of small satellites requires developing a number of novel wireless communication and distributed computing technologies.
In this talk, I will focus on the problem of authenticated communication between the cluster of satellites and external (e.g., other spacecrafts, the ground station, etc.). Our approach extends the conventional threshold signature paradigm by additionally supporting membership changes to the federated system: While traditional systems split the signature key only among an a priori fixed group, our scheme allows evolving membership by repeatedly and securely (re)distributing key shares from the old cluster to the new set of agents. This is realized without resorting to system re-initialization nor relying on a central trusted dealer.
Biography:
Nelly Fazio joined the Department of Computer Science at the City College and the Graduate Center of the City University of New York as an Assistant Professor in September 2008. Since then, she is also a member of the research Center for Algorithms and Interactive Scientific Software (CAISS) at City College. Before joining CUNY, she was a visiting research scientist in the Security group at IBM T.J. Watson Research center, working on security issues of decentralized environments such as mobile ad-hoc networks (MANETs) and sensor networks. Prior to that, she was a postdoctoral researcher in the Content Protection group at IBM Almaden Research Center, where she conducted research on advanced cryptographic key management and tracing technologies.
In this talk, I will focus on the problem of authenticated communication between the cluster of satellites and external (e.g., other spacecrafts, the ground station, etc.). Our approach extends the conventional threshold signature paradigm by additionally supporting membership changes to the federated system: While traditional systems split the signature key only among an a priori fixed group, our scheme allows evolving membership by repeatedly and securely (re)distributing key shares from the old cluster to the new set of agents. This is realized without resorting to system re-initialization nor relying on a central trusted dealer.
Biography:
Nelly Fazio joined the Department of Computer Science at the City College and the Graduate Center of the City University of New York as an Assistant Professor in September 2008. Since then, she is also a member of the research Center for Algorithms and Interactive Scientific Software (CAISS) at City College. Before joining CUNY, she was a visiting research scientist in the Security group at IBM T.J. Watson Research center, working on security issues of decentralized environments such as mobile ad-hoc networks (MANETs) and sensor networks. Prior to that, she was a postdoctoral researcher in the Content Protection group at IBM Almaden Research Center, where she conducted research on advanced cryptographic key management and tracing technologies.
Dr. Fazio's research interests are in cryptography and information security, with a focus on digital content protection. She earned her M.Sc.('03) and Ph.D.('06) in Computer Science from New York University. During her studies, she also conducted research at Stanford University, Ecole Normale Superieure (France) and Aarhus University (Denmark). In 2003, she was awarded the NYU CIMS Sandra Bleistein prize, for "notable achievement by a woman in Applied Mathematics or Computer Science." Her Ph.D. thesis was nominated with honorable mention for the NYU J. Fabri prize, awarded yearly for the "most outstanding dissertation in Computer Science."
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Socially-Aware Recommendations in Collaborative Tagging Sites 23 Nov 2008
Speaker: Sihem Amer-YahiaSenior Research Scientist, Yahoo! Research - NYC
Time: Monday, Nov 17, 2PM
Location: Babbio 221, Stevens Institute of Technology
Host: Wendy Hui Wang
Abstract:
Collaborative tagging sites constitute a unique opportunity for leveraging implicit and explicit social ties in search and recommendations. This talk describes our work in the context of Royal Jelly, a platform for socially-aware recommendations in collaborative tagging sites. The key aspects of RJ are search over recommendations, topic derivation from tags, efficient recommendation algorithms and diversity in recommendations. We describe some of our solutions in that space and an overview of the challenges ahead.
Bio:
Sihem Amer-Yahia is a Senior Research Scientist at Yahoo! Research in NYC. Her interests are at the intersection of databases, search and recommendations in social content sites. Before that, she spent 7 years at AT&T Labs in NJ, working on XML query optimization and XML full-text search. Sihem is an editor of the XML full-text language documents recommended by the W3C. This year, she is on the editorial board of the Springer Encyclopedia of Database Systems as an editor of the XML entry, she also edited the Data Engineering Bulletin special issue on databases, search and recommendations and she is chairing the database track of CIKM 2008.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
PhotoSketch: Photo-Centric Urban 3D Modeling Tool 6 Nov 2008
Prof. George Wolberg
Department of Computer Science
City College of New York
New York, NY 10031
Monday, Nov 10, 2PM
Babbio 221, Stevens Institute of Technology
Department of Computer Science
City College of New York
New York, NY 10031
Monday, Nov 10, 2PM
Babbio 221, Stevens Institute of Technology
Abstract:
Geospatial navigation tools such as Google Earth and Microsoft Virtual Earth are exceedingly popular applications for exploring massive datasets. Their explosive growth provides impetus for photorealistic 3D modeling of urban scenes. Although laser range scanners are traditional sources for detailed 3D models of existing structures, they are prohibitively expensive and generate heavyweight models that are not appropriate for the streaming data that these navigation applications leverage. Instead, lightweight models as produced by photogrammetry tools are better suited for this domain. Unfortunately, photogrammetry requires skilled users for manual camera calibration and complex modeling processes. This makes laser scanning and photogrammetry unsuitable for average consumers equipped with ordinary cameras.
This talk presents the virtues of combining computer vision techniques and photogrammetry to simplify the 3D modeling workflow. The contribution of this work is that it merges the benefits of automatic feature extraction, multiview geometry, an intuitive sketching interface, and dynamic texture mapping to produce lightweight photorealistic 3D models of buildings. Dynamic texture mapping is key to our interactive photo-centric 3D modeling tool whereby the models are edited to best match the projected images. We present results from experiments in urban scenes.
Bio:
Dr. George Wolberg is Professor of Computer Science at the City College of New York / CUNY. He received the B.S. and M.S. degrees in Electrical Engineering from Cooper Union in 1985, and the Ph.D. degree in Computer Science from Columbia University in 1990. He has published widely in image processing, computer graphics, and computer vision journals and conferences, and holds five U.S. patents. He is the author of "Digital Image Warping," (IEEE Computer Society Press, 1990), the first comprehensive monograph on warping and morphing. Prof. Wolberg is the recipient of the Mayor's Award for Excellence in Science and Technology (awarded by Mayor Giuliani in 2000), a CCNY Outstanding Teaching Award, and an NSF Presidential Young Investigator Award. He is a Senior Member of the Institute of Electrical and Electronic Engineers. His research interests include image processing, computer graphics, and computer vision.
Geospatial navigation tools such as Google Earth and Microsoft Virtual Earth are exceedingly popular applications for exploring massive datasets. Their explosive growth provides impetus for photorealistic 3D modeling of urban scenes. Although laser range scanners are traditional sources for detailed 3D models of existing structures, they are prohibitively expensive and generate heavyweight models that are not appropriate for the streaming data that these navigation applications leverage. Instead, lightweight models as produced by photogrammetry tools are better suited for this domain. Unfortunately, photogrammetry requires skilled users for manual camera calibration and complex modeling processes. This makes laser scanning and photogrammetry unsuitable for average consumers equipped with ordinary cameras.
This talk presents the virtues of combining computer vision techniques and photogrammetry to simplify the 3D modeling workflow. The contribution of this work is that it merges the benefits of automatic feature extraction, multiview geometry, an intuitive sketching interface, and dynamic texture mapping to produce lightweight photorealistic 3D models of buildings. Dynamic texture mapping is key to our interactive photo-centric 3D modeling tool whereby the models are edited to best match the projected images. We present results from experiments in urban scenes.
Bio:
Dr. George Wolberg is Professor of Computer Science at the City College of New York / CUNY. He received the B.S. and M.S. degrees in Electrical Engineering from Cooper Union in 1985, and the Ph.D. degree in Computer Science from Columbia University in 1990. He has published widely in image processing, computer graphics, and computer vision journals and conferences, and holds five U.S. patents. He is the author of "Digital Image Warping," (IEEE Computer Society Press, 1990), the first comprehensive monograph on warping and morphing. Prof. Wolberg is the recipient of the Mayor's Award for Excellence in Science and Technology (awarded by Mayor Giuliani in 2000), a CCNY Outstanding Teaching Award, and an NSF Presidential Young Investigator Award. He is a Senior Member of the Institute of Electrical and Electronic Engineers. His research interests include image processing, computer graphics, and computer vision.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Secure Detection of Policy Conflicts in a Multi-Organizational Collaborative Environment 30 Oct 2008
Prof. Basit ShafiqRutgers University
Monday, Nov 3, 2PM
Babbio 221, Stevens Institute of Technology
Abstract
In a collaborative environment, it is often the case that multiple autonomous organizations work together to accomplish a task. However, each collaborating organization may have its own predefined set of business policies, where some of them may be sensitive in nature. For example, as a policy, each organization taking part in the collaboration may have a negative list of organizations that are not allowed to do certain jobs due to conflict of interest or a similar reason. The challenge is to identify organizations capable of accomplishing the task without disclosing the sensitive policies to others and at the same time ensuring adherence to them. While this problem can be trivially solved if one assumes that there exists a trusted entity which has the knowledge of all the policies, we propose an approach that does not rely on a trusted third party. Our proposed approach essentially is capable of detecting policy conflicts, by employing commutative encryption. Through formal proof and analysis, we demonstrate that our approach is both secure and efficient.
BiographyDr. Basit Shafiq is a Research Assistant professor at the Center for Information Management, Integration and Connectivity (CIMIC), Rutgers University. He received his Ph. D. and M.S. in Electrical and Computer Engineering from Purdue University. His research interests include information systems security and privacy, access-control management in distributed systems, Web services composition and verification, ontologies, and distributed multimedia systems. His research work resulted in several publications in well-renowned journals, including, IEEE Transactions on Knowledge and Data Engineering, ACM Transactions on Information and System Security, IEEE Transactions on Multimedia, IEEE Computer, IEEE Communications Magazine, and Journal on Information and Computer Systems. Dr. Shafiq is actively involved in the development of a prototype decision support system for emergency management. This prototype focuses on providing an information sharing environment and a collaboration framework among the various government agencies and private organizations involved in emergency management activities. The research and development work of this project is sponsored by NSF and SAP Labs.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Scaling up Cognitive Radio 25 Oct 2008
Prof. Joe Mitola,Stevens Institute of Technology
Monday, Oct 27, 2pm - 3pm
Babbio 221, Stevens Institute of Technology
Abstract
Cognitive radio technology improves spectrum efficiency via radio etiquettes in which a cognitive radio detects and avoids legacy radios as well as finding and using unoccupied spectrum dynamically. As the complexity of these air interfaces continues to increase, so does the cost of maintaining the related signal processing software. This talk presents results from an investigation into the application of Cartesian genetic programming to evolve signal processing software for signal detection, in lieu of writing the signal processing code manually. The software engineering process includes the reverse engineering of existing software via cognitive linguistics and topological space modeling of the existing and evolving code to achieve the consistency of code modularization needed for genetic programming in lieu of software engineering. Although at an early stage, this approach to software maintenance has some potentially interesting characteristics.
Bio
Dr. Joe Mitola is Distinguished Professor in the Charles V. Schafer School of Engineering and Science and in the School of Systems and Enterprises at Stevens Institute of Technology where his research interests include software defined and cognitive radio and the scaling up of cognition technologies to the enterprise level. Previously, Dr. Mitola was Chief Scientist of the DoD Federally Funded R&D Center of The MITRE Corporation; special assistant to the Director, DARPA; program manager at DARPA; and Chief Scientist of Electronic Systems of E-Systems (now Raytheon). He holds the BS in EE from Northeastern University, the MSE from the Johns Hopkins University; and the Licentiate and Doctorate in Teleinformatics from KTH, The Royal Institute of Technology, Stockholm.
Blinklist | Del.icio.us | Digg | Furl | Ma.gnolia | Newsvine | Spurl | Technorati
Powered by SimplePie 1.0 b3.2, Build 20061124.
