January 30 - CANCELLED
Wednesday - 1:00 PM
4-201 Center for Science and Technology
Prof. Samantha Kleinberg |
Speaker: Prof. Samantha Kleinberg, Assistant Professor of Computer Science at Stevens Institute of Technology
Topic: Inferring Causal Relationships from Large-Scale Time-Series
Abstract: One of the key problems we face with the accumulation of massive datasets (such as electronic health records and stock market data) is the transformation of data to actionable knowledge. In order to use the information gained from analyzing these data to intervene to, say, treat patients or create new fiscal policies, we need to know that the relationships we have inferred are causal. Further, we need to know the time over which the relationship takes place in order to know when to intervene. In this talk I discuss recent methods for finding complex causal relationships and their timing with minimal background knowledge and why inferring the effects of rare causes is both critically important and feasible.
Bio: Samantha Kleinberg is an Assistant Professor of Computer Science at Stevens Institute of Technology. She received her PhD in Computer Science from New York University in 2010 and was a Computing Innovation Fellow at Columbia University in the Department of Biomedical informatics from 2010-2012. Her research centers on developing methods for analyzing large-scale, complex, time-series data. In particular, her work develops methods for finding causes and automatically generating explanations for events, facilitating decision-making using massive datasets. She is the author of Causality, Probability, and Time (Cambridge University Press, 2012).
|
February 20
Wednesday - 1:00 PM
4-201 Center for Science and Technology
Prof. C. Imre Koksal |
Speaker: Prof. C. Imre Koksal, Assistant Professor, Electrical and Computer Engineering, The Ohio State University
Title: Information Theory Enabled Secure Wireless Networking: Scaling Laws and Network Control
Abstract: Current wireless network secrecy mechanisms mainly fall under the umbrella of computational-based cryptography and are vulnerable in scenarios where an adversary has significantly greater computational/processing capabilities. On the other hand, while information theoretic concepts of secrecy are able to provide perfect guarantees, most of the research conducted in this area is limited to physical layer, and (i) ignores issues with packet delays, perhaps the most essential of wireless communication metrics; (ii) does not study the interaction with the key higher layer wireless network functionalities. This talk describes some of our recent efforts to utilize the full power of information-theoretic approaches in developing new solutions to maximize the performance of wireless networks under provable secrecy guarantees.
Bio: C. Emre Koksal received the B.S. degree in electrical engineering from the Middle East Technical University, Ankara, Turkey, in 1996, and the S.M. and Ph.D. degrees in electrical engineering and computer science from Massachusetts Institute of Technology (MIT), Cambridge, in 1998 and 2002, respectively. He was a Postdoctoral Researcher with the Electrical Engineering and Computer Science Department, MIT, and with the School of Communication and Computer Sciences, EPFL, Lausanne, Switzerland, until 2006. Since then, he has been an Assistant Professor with the Electrical and Computer Engineering Department, The Ohio State University (OSU), Columbus. His general areas of interest are wireless communication, communication networks, and information theory. Dr. Koksal is the recipient of the National Science Foundation CAREER Award, the OSU College of Engineering Lumley Research Award, and the co-recipient of an HP Labs–Innovation Research Award, all in 2011. The paper he coauthored was a best student paper candidate in ACM MobiCom 2005. |
February 27
Wednesday - 1:00 PM
4-201 Center for Science and Technology
Prof. Rong Zheng
|
Speaker: Prof. Rong Zheng, Associate Professor, Department of Computer Science, University of Houston
Topic: Sequential Learning in Wireless Monitoring
Abstract: Passive monitoring is a technique where a dedicated set of hardware devices called sniffers, are used to monitor activities in wireless networks. These devices capture transmissions of wireless devices or activities of interference sources in their vicinity, and store packet level or PHY layer information in trace files, which can be analyzed distributively or at a central location. Since most, if not all, infrastructure networks utilize multiple contiguous or non-contiguous channels or bands, an important issue is to determine which set of frequency bands each sniffer operates on to maximize the total amount of information gathered. In this talk, we consider the problem of optimally assigning p sniffers to K channels to monitor the transmission activities in a multi-channel wireless network. The activity of users is initially unknown to the sniffers and is to be learned along with channel assignment decisions. We devise efficient sequential learning approaches and address practical constraints including channel switching time, computation costs and non-stationary network conditions.
Bio: Rong Zheng received her Ph.D. degree from Dept. of Computer Science, University of Illinois at Urbana-Champaign and earned her M.E. and B.E. in Electrical Engineering from Tsinghua University, P.R. China. She is presently an associate professor in the Dept of Computing and Software, McMaster University. She was with the University of Houston from 2004 to 2012.
Rong Zheng's research interests include network monitoring and diagnosis, cyber physical systems, and mobile computing. She is the recipient of the US National Science Foundation CAREER Award in 2006, and University of Houston research excellence award in 2010. She served on the technical program committees of several leading networking conferences, and was the program co-chair of WASA'12 and CPSCom'12.
|
March 8
EECS Seminar
Friday - 1:00 PM
4-201 Center for Science and Technology
Olga Ohrimenko |
Speaker: Olga Ohrimenko, Ph.D. Candidate, Department of Computer Science, Brown University
Title: Privacy and Integrity of Remote Storage and Computation
Abstract: Cloud computing provides on demand remote storage and computation resources at a cost that depends only on the usage rate. However, the loss of the physical control over data and computation raises new challenges in terms of security. These challenges include maintaining privacy of remotely stored data and verification of outsourced computation.
In this talk, we first address the problem of privacy-preserving access to data stored with a cloud provider. Storing the data in encrypted form is a key component in maintaining the privacy of the data. However, encrypting the data is not enough since information about the data may be leaked through the pattern in which users access the data. We show how to achieve efficient privacy-preserving data access using low communication and storage overhead. Our method is based on a combination of encryption, which directly hides data values, and stateless oblivious RAM simulation, which hides the pattern of data accesses. We provide experimental results from our approach and compare it with a more powerful scheme where a client is allowed to keep a state.
In the second part of the talk, we consider the problem of verification of a very common and computationally intensive outsourced computation: keyword search over a document collection. We present a method that allows a client to verify that the result she receives to her query is complete and sound. We propose a solution that requires the cloud provider to compute not only the result, but also a cryptographic proof of the computation. The client then uses the proof to verify that the list of documents returned is correct w.r.t. her query and document collection. We show that our solution adds a small overhead: the size of the proof is as large as the query result, and the verification time for the client is negligible and takes milliseconds.
Bio: Olga Ohrimenko is a Ph.D. candidate in the Department of Computer Science at Brown University working with Professor Roberto Tamassia. Her interests include devising methods that address privacy, integrity and security issues that emerge in the cloud computing environment.
Olga received a B.CS. (Hons) degree from The University of Melbourne in 2007 and her M.Sc. degree from Brown University in 2010. She was also an intern at Google, Microsoft Research and IBM Research. |
March 20
Wednesday - 1:00 PM
4-201 Center for Science and Technology
Prof. Yuejie Chi |
Speaker: Prof. Yuejie Chi, Assistant Professor, Department of Electrical and Computer Engineering, Department of Biomedical Informatics, The Ohio State University
Abstract: The paper studies the problem of recovering a spectrally sparse object from a small number of time domain samples. Specifically, the object of interest with ambient dimension $n$ is assumed to be a mixture of $r$ complex sinusoids, while the underlying frequencies can be continuous-valued on the unit disk of multi-dimension. This problem arises in many applications such as radar, astrophysics, microscopy imaging, etc. Conventional compressed sensing paradigms suffer from the basis mismatch issue when imposing a discrete dictionary on the Fourier representation. To address this problem, we develop a novel nonparametric algorithm, called enhanced matrix completion (EMaC), based on structured matrix completion. The algorithm starts by arranging the data into a low-rank enhanced form with k-fold Hankel structure, then attempts recovery via nuclear norm minimization. Under mild incoherence conditions, EMaC allows perfect recovery as soon as the number of samples exceeds the order of $O(r log^{2} n)$. We also show that, in many instances, accurate completion of a low-rank k-fold Hankel matrix is possible when the number of observed entries is proportional to the information theoretical limits (except for a logarithmic gap). We further provide theoretical guarantees that EMaC is robust to noise and sparse corruptions. Finally, the performance of EMaC and its applicability to super resolution are demonstrated by numerical experiments.
Bio: Yuejie Chi is currently an assistant professor at Ohio State University in a joint role with electrical and computer engineering and biomedical informatics. She obtained her Ph.D. from Princeton University under the supervision of Prof. Robert Calderbank in 2012. In addition to conducting research at Princeton, Chi has been a visiting scholar at Stanford University, Duke University and Colorado State University. She also completed research internships at Mitsubishi Electric Research Lab and Qualcomm, Inc., where she was awarded a prestigious Roberto Padovani Scholarship. She received a best paper award from the 2012 International Conference on Acoustics, Speech, and Signal Processing (ICASSP). Chi’s research interests nclude statistical signal processing, compressive sensing, machine learning, wireless communications and networks, and bioinformatics. |
March 22
EECS Seminar
Friday - 1:00 PM
4-201 Center for Science and Technology
Prof. Mahesh Tripunitara
|
Speaker: Prof. Mahesh Tripunitara, Department of Electrical and Computer Engineering, University of Waterloo-Canada
Title: An attack-and a defence-mechanism in the context of hardware security
Abstract: With the increased outsourcing of the fabrication of digital Integrated Circuits (ICs), security is seen as a concern. The threat agent is someone at the foundry, perhaps in collusion with a designer or a user, that maliciously modifies ICs during fabrication, for example, by inserting a backdoor. In this talk, I will discuss two pieces of on-going work in this context.
The first piece is the realization and validation of a non-deterministic hardware timer that can be used to trigger a backdoor. Prior work has considered deterministic timers, i.e., those that go off with probability 1, and has left open issues regarding the effectiveness, from the standpoint an attacker, of non-deterministic timers, i.e., those that have a random component. Our work addresses these open issues and shows that such timers can be realized with powerful properties to an attacker, in a manner that the bar on potential defence mechanisms is raised considerably.
The second piece of work I will discuss is a defence-mechanism that leverages 3D IC technology that splits a circuit into multiple tiers, each of which may be fabricated separately, and then stacked vertically and connected using Through-Silicon Vias (TSVs). Prior work has proposed that such technology can be used to secure digital ICs, but provides no technical insight or details on how this would work. I will discuss our work that proposes a concrete way of leveraging such technology for security. This includes a characterization of security, and the computational complexity of the underlying problem. I will discuss also an approach we have implemented and present empirical results on benchmark circuits, and a case-study of a circuit for DES.
(This is joint work with Frank Imeson and Siddharth Garg of the University of Waterloo.)
Bio: Mahesh Tripunitara is an assistant professor in the ECE department at the University of Waterloo in Canada, where he had been since 2009. He works mostly in information security, on problems in access control, conditional payments, cryptographic key transport and more recently, computer hardware. He has a PhD in computer science from Purdue University, and about 9 years of industry-experience. |
March 27
Wednesday - 1:00 PM
4-201 Center for Science and Technology
Prof. Nitin Vaidya
|
Speaker: Prof. Nitin Vaidya, Department of Electrical and Computer Engineering, University of Illinois, Urbana-Champaign
Title: Resilient Distributed Consensus
Abstract: Consensus algorithms allow a set of nodes to reach an agreement on a quantity of interest. For instance, a consensus algorithm may be used to allow a network of sensors to determine the average value of samples collected by the different sensors. Similarly, a consensus algorithm can also be used by the nodes to synchronize their clocks. Research on consensus algorithms has a long history, with contributions from different research communities, including distributed computing, control systems, and social science.
In this talk, we will discuss two resilient consensus algorithms that can perform correctly despite the following two types of adversities: (i) In wireless networks, transmissions are subject to transmission errors, resulting in packet losses. We will discuss how "average consensus" can be achieved over such lossy links, without explicitly making the links reliable, for instance, via retransmissions. (ii) In a distributed setting, some of the nodes in the network may fail or may be compromised. We will discuss a consensus algorithm that can tolerate "Byzantine" failures in partially connected networks.
Bio: Nitin Vaidya is a Professor of Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign. His research interests span distributed algorithms, fault-tolerant computing, and wireless networks. Nitin has held visiting positions at Technicolor Paris Lab, TU-Berlin, IIT-Bombay, Microsoft Research-Redmond, and Sun Microsystems, as well as a faculty position at the Texas A&M University. He has co-authored papers that received awards at several conferences, including 2007 ACM MobiHoc and 1998 ACM MobiCom. He has served as the Editor-in-Chief for the IEEE Transactions on Mobile Computing. Nitin is a Fellow of the IEEE. For more information, please visit http://users.crhc.illinois.edu/nhv. |
April 3
Wednesday - 1:00 PM
4-201 Center for Science and Technology
Prof. Samantha Kleinberg |
Speaker: Prof. Samantha Kleinberg, Assistant Professor of Computer Science at Stevens Institute of Technology
Topic: Inferring Causal Relationships from Large-Scale Time-Series
Abstract: One of the key problems we face with the accumulation of massive datasets (such as electronic health records and stock market data) is the transformation of data to actionable knowledge. In order to use the information gained from analyzing these data to intervene to, say, treat patients or create new fiscal policies, we need to know that the relationships we have inferred are causal. Further, we need to know the time over which the relationship takes place in order to know when to intervene. In this talk I discuss recent methods for finding complex causal relationships and their timing with minimal background knowledge and why inferring the effects of rare causes is both critically important and feasible.
Bio: Samantha Kleinberg is an Assistant Professor of Computer Science at Stevens Institute of Technology. She received her PhD in Computer Science from New York University in 2010 and was a Computing Innovation Fellow at Columbia University in the Department of Biomedical informatics from 2010-2012. Her research centers on developing methods for analyzing large-scale, complex, time-series data. In particular, her work develops methods for finding causes and automatically generating explanations for events, facilitating decision-making using massive datasets. She is the author of Causality, Probability, and Time (Cambridge University Press, 2012).
|
April 5
IEEE Distinguished Lecturer
Friday - 1:00 PM
4-201 Center for Science and Technology
Prof. Narayan B. Mandayam
|
Speaker: Prof. Narayan B. Mandayam, WINLAB, Dept. of Electrical & Computer Engineering, Rutgers University
Title: Network Coding as a Dynamical System
Abstract: In the last few years the area of network coding has seen an explosive growth in research activity while being touted as the foundation on which several applications related to the robust operation of both wired and wireless networks can be built. The breadth of areas that have been touched by network coding is vast and incudes not only the traditional disciplines of information theory, coding theory and networking, but also topics such as algorithms, combinatorics, distributed storage, network monitoring, content delivery, and security. There has been a rich outpouring of theoretical results for network coding in wireless networks as well as an equally excellent body of work that discusses implementation issues. In spite of all this excellent progress in this rich area, what has been missing is a simple framework that can allow explaining the evolution of network coding in an arbitrary wireless network. For example, given an arbitrary wireless network and a network coding strategy, a question that remains to be answered is how the rank/state of the nodes in the network evolves over time. Further, if there are changes in the underlying wireless network either through changes in the PHY layer, MAC layer or due to other factors such as mobility or traffic, how does this impact the evolution of network coding over this arbitrary network? Answering such questions is of paramount importance for network practitioners and can provide further insights into making network coding a tool closer to the design of real wireless networks. In this talk I will outline a framework based on differential equations that allows modeling of dynamics of wireless network coding and enables the design of cross-layer radio resources allocation algorithms.
Bio: Narayan B. Mandayam is currently the Peter D. Cherasia Fraculty Scholar at Rutgers University, where he also serves as Associate Director of WINLAB. His research interests are in various aspects of wireless data transmission including system modeling and performance, signal processing and radio resource management with emphasis on techniques for cognitive radio networks. Dr. Mandayam is a recipient of the Fred W. Ellersick Prize from the IEEE Communications Society in 2009 for his work on dynamic spectrum access models and spectrum policy. He is also a recipient of the Institute Silver Medal from the Indian Institute of Technology in 1989 and the National Science Foundation CAREER Award in 1998. He is a coauthor of the books: Principles of Cognitive Radio (Cambridge University Press, 20012) and Wireless Networks: Multiuser Detection in Cross-Layer Design (Springer, 2005). He has served as an Editor for the journals IEEE Communication Letters and IEEE Transactions on Wireless Communications. He has served as a guest editor of the IEEE JSAC Special Issues on Adaptive, Spectrum Agile and Cognitive Radio Networks (2007) and Game Theory in Communication Systems (2008). He is a Fellow of the IEEE and currently serves as a Distinguished Lecturer of the IEEE Communications Society. |
April 19
Friday - 10:00 AM
106 Lundgren Room – Life Sciences Complex
Dr. J. R. Rao |
Speaker: Dr. J. R. Rao, Director, Security Research and Member, IBM Academy of Technology, IBM Thomas J. Watson Research Center, Yorktown Heights, NY
Title: CyberSecurity Research at IBM
Abstract: Over the last decade, the problem of CyberSecurity has become the most pressing and challenging security problem of our time. As CyberSecurity threats and attacks grow in frequency and the enterprise risk escalates rapidly, businesses and governments are searching for innovative technologies to combat the problem more effectively. Enterprise customers today have deployed numerous security controls including security sensors such as intrusion prevention and detection systems as well as security tooling for identity, access and audit management. These systems enable enterprises to manage their security posture, generating a multitude of event alert streams as well as logs and audit records that contain potentially actionable intelligence that today is typically not fully mined nor available in real-time.
In this talk, we describe a five-pronged research agenda for equipping enterprises with a new approach to combating CyberSecurity challenges. The first initiative focuses on identifying critical high-value business assets in the enterprise and discovering where they exist in a scalable manner. The second initiative enables enterprises to comprehensively monitor and build behavioral models of how such assets are accessed over networks, devices, user, social networks, applications and business processes with the goal of detecting suspicious activity to provide insights for superior risk management. Once the risk to high value assets has been assessed, the third initiative focuses on building innovative security technologies that can be used to reinforce the security of the platform and systems on which such assets are hosted. While the first three initiatives focus on securing high value assets, endpoints, especially mobile endpoints, remain the weak link in the chain. The fourth initiative focuses on achieving end-to-end security by focusing on securing end-to-end security. The final initiative focuses on exploring state of the art techniques in cryptography, privacy and secure engineering to achieve security by design.
Bio: J.R. Rao is Director for Security Research at IBM. Housed in IBM's global labs, the team comprises over a hundred researchers who work in the areas of Cryptography, CyberSecurity, Cloud and Mobile Security and Secure Platform Technologies. JR works closely with customers, academic partners and IBM business units to drive new and innovative technologies into IBM's products and services and definitive industry leading standards, JR has published widely in premier security conferences and workshops and holds numerous US and European patents. He is a member of the Industry Advisory Board of the Georgia Tech Information Security Center, is a member of the prestigious IBM Academy of Technology as well as member-emeritus of the IFIP's Working Group 2.3 (Programming Methodology). JR obtained his doctorate degree from the University of Texas at Austin in 1992, a master's degree from the State University of New York at Stony Brook in 1986 and a bachelor's degree from the Indian Institute of Technology, Kanpur in 1984. |
April 24
Wednesday - 1:00 PM
4-201 Center for Science and Technology
Prof. Lanne Hirshfield |
Speaker: Prof. Leanne Hirshfield, Associate Research Professor
Syracuse University Newhouse School
Title: Using non-invasive brain measurement to benefit the cyber-security domain
Abstract: Dr. Hirshfield runs a DURIP funded lab with $468K of cutting edge non-invasive sensor equipment. This equipment includes Hitachi Medical’s 52-channel near-infrared spectroscopy system (fNIRS), a 10-channel B-alert wireless EEG, Affectiva’s wireless Galvanic Skin Response (GSR) sensor, FaceLAB’s deskmounted eyetracker, and accompanying Morae usability software. With this lab, Dr. Hirshfield has used fNIRS and machine learning to objectively measure computer users’ changing levels of cognitive load while their computer systems were disrupted by a range of subtle manipulations (i.e., slowed keypress rate, dropping keystrokes, causing internet pop ups to occur). Dr. Hirshfield recently broadened her research to measure the states of trust and suspicion using fNIRS and GSR. In this ongoing research, she merges domain knowledge from the neuroscience field with custom machine learning algorithms to predict, in real-time, the changing levels of trust and suspicion experienced by a computer user over time. In this talk, Dr. Hirshfield will describe results from recent experiments where various aspects of trust and suspicion were manipulated and measured. She’ll also place the results in context, explaining why the accurate, objective, and real-time prediction of trust and suspicion with physiological sensors has the potential to greatly benefit the cyber domain.
Bio: Dr. Leanne Hirshfield is an Associate Research Professor at Syracuse Univerity’s Newhouse School. Dr. Hirshfield received her Ph.D.in Computer Science from Tufts University in 2009 having completed her undergraduate and M.S. degrees in Computer Science at Hamilton College and the Colorado School of Mines, respectively. Her specialty is in the area of Human-Computer Interaction (HCI) and she has a wealth of experience designing and evaluating various user interfaces. Dr. Hirshfield’s research explores the use of non-invasive brain measurement to passively classify user states in order to enhance usability testing and adaptive system design. Dr. Hirshfield’s research focuses on the use of a relatively new, non-invasive brain imaging device called functional near-infrared spectroscopy (fNIRS). The fNIRS device is safe, portable, robust to noise, and it can be implemented wirelessly; making it ideal for research in HCI. While much of her research focuses on fNIRS, Dr. Hirshfield also works extensively with EEG, galvanic skin response sensors, and eyetracking devices in her lab. In addition to her basic research, Dr. Hirshfield oversees military-oriented research efforts that focus on measuring and predicting trust, suspicion, and situational awareness using the lab’s sensors.
|