Select a calendar:
Filter March Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for March
-
Rich Representations for Detailed Visual Recognition
Mon, Mar 04, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Subhransu Maji, TTI Chicago
Talk Title: CS Colloquium: Subhransu Maji (TTI Chicago)
Series: CS Colloquium
Abstract: As humans, we have the remarkable ability to perceive the world around us in minute detail -- we can estimate material and metric properties of objects, localize people in images, describe what they are doing, and even identify them! Despite many successes, computational models for reliably inferring such fine-grained properties from images are lacking. I'll describe our attempts in developing models for such detailed recognition by improving the interplay of data, computation and humans in three ways.
First, I'll present computationally efficient classifiers for visual recognition which is a key ingredient of many recognition systems. These approximate non-linear kernel SVM classifiers that are widely used in computer vision, while being exponentially faster during training and testing, making them practical for large-scale recognition tasks or detection. Second, I'll show how humans can enable better and interpretable models for detailed recognition. Visual categories are decomposed using novel parts called "poselets" that are semantically aligned to human annotations. These provide a basis for high-level recognition and lead to simple, accurate and interpretable architectures for learning and recognition. The proposed models rely on annotations of landmarks and attributes during learning. Unfortunately, deciding on the right set of landmarks or attributes to annotate can be a challenging task. In the third part of the talk, I'll present a relative annotation framework that overcomes some of the fallacies of traditional annotation methods, while enabling discovery of rich semantic structure within the category when the set of the annotation labels are not known ahead of time. I'll present experiments on semantic part and attribute discovery for visually diverse categories such as buildings, airplanes and birds.
Biography: Subhransu Maji received the BTech degree in computer science and engineering from the Indian Institute of Technology, Kanpur, in 2006, and the PhD degree in computer science from the University of California, at Berkeley, in 2011. He is currently a research assistant professor at TTI Chicago. Earlier he was an intern in Google's image search group and INRIA's LEAR group, and a visiting researcher at Microsoft Research India and the CLSP center at Johns Hopkins University. He received the medal for the best graduating student in the computer science department from IIT Kanpur. He was one of the recipients of the Google graduate fellowship in 2008 and a best paper award at ICIF 2009. His primary interests are in computer vision and machine learning, with focus on representations and efficient algorithms for rich visual recognition.
Host: Gerard Medioni
Location: Hedco Neurosciences Building (HNB) - 100
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Hierarchical processing and the neurobiology of language
Tue, Mar 05, 2013 @ 02:00 PM - 03:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Ina Bornkessel-Schlesewsky and Matthias Schlesewsky, Department of Germanic Linguistics, University of Marburg, and Department of English and Linguistics, Johannes Gutenberg-University Mainz
Talk Title: Hierarchical processing and the neurobiology of language
Abstract: Hierarchical processing has been posited as a basic property of neurobiological organisation both in the visual (e.g. Felleman & Van Essen, 1991) and auditory (Rauschecker, 1998) systems. It is also an important characteristic of a recent neurobiological model of speech processing (Rauschecker & Scott, 2009), which builds upon insights from the auditory system of non-human primates. By contrast, long-standing neurocognitive assumptions about the organisation of language in the brain (e.g. the notion that Broca's region in left frontal cortex is crucial for grammatical processing) are often incompatible with the tenet of hierarchical processing. Here, we outline a new neurobiological approach to language processing which applies the principle of hierarchical organisation to sentence and discourse comprehension (Bornkessel- Schlesewsky & Schlesewsky, in press). We show how the architectural consequences of this basic design principle help to reconcile a number of theoretical and empirical puzzles within the existing literature on the neuroscience of language. Furthermore, they lead to novel and sometimes surprising hypotheses (e.g. regarding the neural bases for structuring sentences in time).
Biography: Ina Bornkessel-Schlesewsky
Department of Germanic Linguistics, University of Marburg
and Matthias Schlesewsky
Department of English and Linguistics, Johannes Gutenberg-University Mainz
Host: Michael Arbib
Location: Ray R. Irani Hall (RRI) - 101
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Big Data Analytics with Parallel Jobs
Tue, Mar 05, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Ganesh Ananthanarayanan, UC Berkeley
Talk Title: CS Colloquium: Ganesh Ananthanarayanan (UC Berkeley)
Series: CS Colloquium
Abstract: Extensive data analysis has become the enabler for diagnostics and decision making in many modern systems. These analyses have both competitive as well as social benefits. To cope with the deluge in data that is growing faster than Mooreââ¬â¢s law, computation frameworks have resorted to massive parallelization of analytics jobs into many fine-grained tasks. These frameworks promised to provide efficient and fault-tolerant execution of these tasks. However, meeting this promise in clusters spanning hundreds of thousands of machines is challenging and a key departure from earlier work on parallel computing.
A simple but key aspect of parallel jobs is the all-or-nothing property: unless all tasks of a job are provided equal improvement, there is no speedup in the completion of the job. This talk will demonstrate how the all-or-nothing property impacts replacement algorithms in distributed caches for parallel jobs. Our coordinated caching system, PACMan, makes global caching decisions and employs a provably optimal cache replacement algorithm. A highlight of our evaluation using workloads from Facebook and Bing datacenters is that PACManââ¬â¢s replacement algorithm outperforms even Beladyââ¬â¢s MIN (that uses an oracle) in speeding up jobs. Along the way, I will also describe how we broke the myth of disk-localityââ¬â¢s importance in datacenter computing and solutions to mitigate straggler tasks.
Biography: Ganesh Ananthanarayanan is a PhD candidate in the University of California at Berkeley, working with Prof. Ion Stoica in the AMP Lab. His research interests are in systems and networking, with a focus on cloud computing and large scale data analytics systems. Prior to joining Berkeley, he worked for two years at Microsoft Researchââ¬â¢s Bangalore office. More details about Ganeshââ¬â¢s work can be found here: http://www.cs.berkeley.edu/~ganesha/.
Host: Ramesh Govindan
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Bootstrapping Vehicles: a Formal Approach to Unsupervised Sensorimotor Learning Based on Invariance
Thu, Mar 07, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Andrea Censi, Caltech
Talk Title: CS Colloquium: Andrea Censi (CalTech)
Series: CS Colloquium
Abstract: Imagine you are a brain that wakes up in an unknown (robotic) body. You are connected to two streams of uninterpreted observations and commands. You have zero prior information on the body morphology, its sensors, its actuators, and the external world. Would you be able to "bootstrap" a model of your body from scratch, in an unsupervised manner, and use it to perform useful tasks? This bootstrapping problem sits at the intersection of numerous scientific questions and engineering problems. Biology gives us a proof of existence of a solution, given that the neocortex demonstrates similar abilities.
I am interested in understanding whether the bootstrapping problem can be formalized to the point where it can be solved with the rigour of control theory. I will discuss a tractable subset of the set of all robots called the "Vehicles Universe", which I consider a pimped-up version, with modern sensors, of Braitenberg's Vehicles. I will show that the dynamics of three "canonical" robotic sensors (camera, range-finder, field sampler) are very similar at the "sensel" level. I will present classes of models that can capture the dynamics of those sensors simultaneously and allow exactly the same agent to perform equivalent spatial tasks when embodied in different robots. I will discuss immediate applications to intrinsic sensor calibration and fault detection.
A key concern of mine is to precisely characterize the "assumptions" of the agents. I will show that assumptions regarding the representation of the data can be described by the largest group of transformations on observations/commands to which the agent behavior is invariant. This suggests that one of the basic concerns of a bootstrapping agent is being able to reject these "representation nuisances".
Reference: the homonymous dissertation, available at http://purl.org/censi/2012/phd
Biography: Andrea Censi is a postdoctoral scholar in Computing and Mathematical Sciences at the California Institute of Technology. He received the Laurea and Laurea Specialistica degrees (summa cum laude) in control engineering and robotics from Sapienza University of Rome, Italy, in 2005 and 2007, respectively, and a Ph.D. in Control & Dynamical Systems from the California Institute of Technology in 2012. He is broadly interested in perception and decision making problems for natural and artificial embodied agents, and in particular in estimation, filtering, and learning in robotics.
Website: http://andrea.caltech.edu/
Host: Fei Sha
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Wyatt Lloyd (Princeton): Stronger Consistency and Semantics for Geo-Replicated Storage
Mon, Mar 11, 2013 @ 04:00 PM - 05:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Wyatt Lloyd, Princeton
Talk Title: Stronger Consistency and Semantics for Geo-Replicated Storage
Series: CS Colloquium
Abstract: Geo-replicated storage systems provide the backend for massive-scale websites such as Google and Facebook. These storage systems seek to provide an always-on experience where every operation completes quickly because of a widely demonstrated link between page load times, user engagement, and revenue. We term systems that provide an always-on experience and can handle data at the required scale "ALPS"
systems because they provide four key properties: Availability, Low latency, Partition tolerance, and Scalability.
Previous ALPS systems made large usability sacrifices in pursuit of their scale and performance goals. They settled for eventually consistent replication between datacenters and inconsistent batch operations within them. My research shows that these sacrifices are not fundamental.
In this talk, I will present the first ALPS system to provide consistency that is stronger than eventual. Specifically, I will show how to provide causal consistency for data stored in multiple datacenters, each of which spreads the data across many servers. Then, I will show how to strengthen the semantics of that system with a richer data model as well as read-only and write-only transactions.
Biography: Wyatt Lloyd is a Ph.D. candidate in Computer Science at Princeton University. His research interests include the distributed systems and networking problems that underlie the architecture of large-scale websites, cloud computing, and big data. He received his masters degree in Computer Science from Princeton University, and a bachelors degree in Computer Science from Penn State University.
Host: Ethan Katz-Bassett
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Barzan Mozafari (CSAIL MIT): Building the Next Generation of Data-Intensive Systems: From Complex Event Processing to Large-Scale Analytics
Tue, Mar 12, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Barzan Mozafari, CSAIL MIT
Talk Title: Building the Next Generation of Data-Intensive Systems: From Complex Event Processing to Large-Scale Analytics
Series: CS Colloquium
Abstract: Databases have been a successful abstraction for accessing and managing data in traditional workloads. However, the rapid growth of data and the demand for more complex analytics have significantly hindered the scalability and applicability of these systems beyond classic business data processing scenarios. In my talk, I will explain how my research addresses these two challenges. First, I will introduce a system that I have built for supporting complex event processing over both stored and streaming data. This system extends existing database query languages with minimal but powerful constructs that enable a wide range of advanced applications, such as high-frequency trading, click-stream analysis, and the analysis of function-call traces. Using the recently proposed Visibly Pushdown Automata as the underlying model of this system, I will present several optimization techniques for efficient implementation of these languages, leading to higher throughput than its predecessors by several orders of magnitude. In the second part of my talk, I will turn to the scalability challenges, and briefly introduce a parallel query engine called BlinkDB that enables interactive, ad-hoc queries over massive volumes of data in a MapReduce cluster. I will demonstrate how BlinkDB employs sophisticated optimization and sampling strategies to achieve sub-second latency on tens of terabytes to petabytes of data.
Biography: Barzan Mozafari is currently a Postdoctoral Associate at Massachusetts Institute of Technology. He earned his PhD in Computer Science from the University of California at Los Angeles. He is passionate about building practical, large-scale data-intensive systems, with a particular interest in database-as-a-service, distributed systems, and the integration of machine learning and crowdsourcing into database systems. He has received several fellowships and awards, including SIGMOD 2012's best paper award for his work on high-performance complex event processing.
Host: Cyrus Shahabi
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Dennis Shasha (Courant Institute - NYU)
Wed, Mar 13, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dennis Shasha , NYU
Talk Title: Storing Clocked Programs Inside DNA: A Simplifying Framework for Nanocomputing
Series: CS Colloquium
Abstract: In the history of modern computation, large mechanical calculators preceded computers. A person would sit there punching keys according to a procedure and a number would eventually appear. Once calculators became fast enough, it became obvious that the critical path was the punching rather than the calculation itself. That is what made the stored program concept vital to further progress. Once the instructions were stored in the machine, the entire computation could run at the speed of the machine.
This work shows how to do the same thing for DNA computing.
Rather than asking a robot or a person to pour in specific strands at different times in order to cause a DNA computation to occur (by analogy to a person punching numbers and operations into a mechanical calculator), the DNA instructions are stored within the solution and guide the entire computation. We show how to store straight line programs, conditionals, loops, and a rudimentary form of subroutines. We propose a novel machine motif which constitutes an instruction stack, allowing for the clocked release of an arbitrary sequence of DNA instruction or data strands. The clock mechanism is built of special strands of DNA called "tick" and "tock." Each time a "tick" and "tock" enter a DNA solution, a strand is released from an instruction stack (by analogy to the way in which as a clock cycle in an electronic computer causes a new instruction to enter a processing unit). As long as there remain strands on the stack, the next cycle will release a new instruction strand. Regardless of the actual strand or component to be released at any particular clock step, the "tick" and "tock" fuel strands remain the same, thus shifting the burden of work away from the end user of a machine and easing operation. Pre-loaded stacks enable the concept of a stored program to be realized as a physical DNA mechanism.
We demonstrate by a series of experiments conducted in Ned Seeman's lab that it is possible to "initialize" a clocked stored program DNA machine. We end with a discussion of the design features of a programming language for clocked DNA programming. There is a lot left to do.
Biography: Dennis Shasha is a professor of computer science at the Courant Institute of New York University where he works with biologists on pattern discovery for network inference; with computational chemists on algorithms for protein design; with physicists and financial people on algorithms for time series; on clocked computation for DNA computing; and on computational reproducibility. Other areas of interest include database tuning as well as tree and graph matching. Because he likes to type, he has written six books of puzzles about a mathematical detective named Dr. Ecco, a biography about great computer scientists, and a book about the future of computing. He has also written five technical books about database tuning, biological pattern recognition, time series, DNA computing, statistics, and causal inference in molecular networks. He has written the puzzle column for various publications including Scientific American.
Host: Shahram Ghandeharizadeh
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Maya Cakmak (Willow Garage): Enabling End-Users to Program General-Purpose Robots
Thu, Mar 14, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Maya Cakmak, Willow Garage
Talk Title: CS Colloquium: Maya Cakmak (Willow Garage): Enabling End-Users to Program General-Purpose Robots
Series: CS Colloquium
Abstract: Advances in robotics research and supporting technologies are enabling the development of more and more sophisticated general-purpose robots. These robots could carry out useful tasks for and with humans, in application domains ranging from elder care to manufacturing. Given the diversity of the operation environments and user needs for such robots, it is infeasible to fully program them prior to their deployment. Therefore, a key challenge is to enable end-users to program a general-purpose robot for their own unique purposes. In this talk, I will highlight some of the challenges in allowing everyday people, who have no prior experience with robots or programming, to teach new skills or tasks to a robot. I will present techniques that I have developed for addressing those challenges, focusing on mechanisms for the robot to ask questions to its user. I will demonstrate how user-studies involving real human-robot interactions can lead to alternative representations, algorithms and user-interfaces that improve the efficiency and effectiveness of the robot, as well as the user experience. I will conclude with a research agenda towards long-term deployment of end-user programmable robots.
Biography: Maya Cakmak is a post-doctoral research fellow at Willow Garage. She received her Ph.D. in Robotics from the Georgia Institute of Technology in 2012. Her research interests are at the intersection of Human-Robot Interaction and Programming by Demonstration. In particular, her research aims to develop functionalities and interfaces for personal robots that can be programmed by their end-users to assist everyday tasks. Maya's work has been published at major Robotics and AI conferences and journals, demonstrated live in various venues and has been featured in numerous media outlets.
Host: Fei Sha
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Dr Yael Moses (IDC Israel): Photo-Sequencing
Tue, Mar 19, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Yael Moses, IDC Israel
Talk Title: Dr Yael Moses (IDC Israel): Photo-Sequencing
Series: CS Colloquium
Abstract: Dynamic events such as family gatherings, concerts or sports events are often captured by a group of people. The set of still images obtained this way is rich in dynamic content but lacks accurate temporal information. We propose a method for photo-sequencing -- temporally ordering a set of still images taken asynchronously by a set of uncalibrated cameras. Photo-sequencing is an essential tool in analyzing (or visualizing) a dynamic scene captured by still images. We demonstrate successful photo sequencing on several challenging collections of images taken using a number of mobile phones.
This is a joint work with Tali Basha and Shai Avidan from TAU.
Biography: Yael Moses received a Ph.D. in computer science from the Weizmann Institute, Israel. She is currently a senior lecturer in the EﬠArazi School of Computer Science at the Interdisciplinary Center (IDC), Herzliya. Her early work concentrated on theoretical aspects of object recognition. Recently, she has been focusing on various aspects of multi-camera systems and distributed computer vision.
Host: Gerard Medioni
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Huijia Rachel Lin (Boston U): Concurrent Security
Mon, Mar 25, 2013 @ 04:00 PM - 05:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Huijia Rachel Lin, Boston U, CSAIL, MIT
Talk Title: Concurrent Security
Series: CS Colloquium
Abstract: Cryptographic protocols have been developed for a variety of tasks, including electronic auctions, electronic voting systems, privacy preserving data mining and more. Traditionally, these cryptographic protocols were analyzed in a simple ââ¬Åstand-aloneââ¬Â model which considers a single execution of the protocols taking place in isolation. Yet, in open networks, such as the Internet, executions of cryptographic protocols may occur concurrently. This concurrency undermines the security of protocols designed for the simple ââ¬Åstand-aloneââ¬Â model. As a consequence, in the past two decades, the study of concurrent security has been a main effort in Cryptography.
In this talk, I will present the first concurrently-secure protocols that enable securely performing general tasks (including all the above-mentioned ones), without relying on any trusted infrastructures or strong hardness assumptions. In particular, I introduce a novel technique that transforms any cryptographic protocol designed for the simple "stand-alone" setting, into one that is secure under concurrent executions. On the way, I solve a two-decade-old open problem, originating in the seminal paper introducing concurrent security.
Biography: Huijia Lin is currently a post-doctoral researcher at the MIT Computer Science and Artificial Intelligence Laboratory and the Department of Computer Science at Boston University. Earlier, she obtained a PhD in Computer Science from Cornell University. Her research interests are in the field of Cryptography.
Host: Shaddin Dughmi
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Dom Massaro: Technology Assisted Reading Acquisition (TARA): Children Acquiring Literacy Naturally
Tue, Mar 26, 2013 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dom Massaro, University of California, Santa Cruz
Talk Title: Technology Assisted Reading Acquisition (TARA): Children Acquiring Literacy Naturally
Series: ICT Distinguished Lecture
Abstract: Society faces increasing challenges in the ability to support the infrastructure of a literate world. Virtual teachers, the internet, and the ceaseless access to information hold promise. To date, however, these potential solutions do not consider research in cognitive science and the potential of the learning brain. As background, the talk reviews our previous research, technology, and applications in speech perception and language learning using our computer-animated face, Baldi. Included is a project to enhance the ability of hearing-challenged and deaf persons to understand conversational speech in face-to-face spoken interactions. The talk offers the possibility of how universal literacy can be achieved with minimal cost, allowing a revolutionary new age that challenges the survival of our educational institutions and society as we know them. It questions the commonly held belief that written language requires formal instruction and schooling whereas spoken language is seamlessly acquired from birth onward by natural interactions with persons who talk. The objectives are to prototype physical systems that exploit developments in behavioral science and technology to a) automatically recognize speech, objects, and actions and b) to display corresponding written descriptions. The goal is to create an interactive system TARA to allow infants, toddlers, and preschool children to acquire literacy naturally.
http://mambo.ucsc.edu/people/dominic-massaro.html
Psyentific Mind
http://psyentificmind.com/
Biography: Dom Massaro is currently a Research Professor at the University of California, Santa Cruz and has had an extended career of innovative language research with preschool and school children as well as adults. Dom has researched both reading and speech perception for four decades, and has advanced these fields empirically, theoretically, and technologically. He also has valuable experience of applying technology and behavioral science to real-world problems. He invented Kid Klok, an educational easy-to-read analog clock, available in both physical and software embodiments. Based on his scientific scholarship and his concomitant development of technology, he co-founded several companies which developed successful products for language learning for language-challenged children such as those with hearing loss and autism. Dom is currently president of Psyentific Mind, a company aimed at using technology and psychology to expand the reach of the human mind. His current focus is Technology Assisted Reading Acquisition (TARA).
Massaro (1989). Child's Easy-To-Read Timepiece. United Startes Patent Number 4,885,731. December 5,1989.
https://itunes.apple.com/us/app/kid-klok/id461743662?mt=8
Massaro, D. W. (1998). Perceiving Talking Faces: From speech perception to a behavioral principle, Cambridge, MA: MIT Press.
Massaro, D. W. (2011). Method And System For Acquisition Of Literacy. Patent Application Number 13/253,335, October 5, 2011. http://www.google.com/patents?id=AwAMAgAAEBAJ&printsec=frontcover&dq=13/253,335&hl=en&sa=X&ei=cz7fT4jEEqS42QXTwOmuDQ&sqi=2&ved=0CDUQ6AEwAA
Massaro, D. W. (2012). Acquiring Literacy Naturally: Behavioral science and technology could empower preschool children to learn to read naturally without instruction. American Scientist, 100, 324-333. http://mambo.ucsc.edu/wp-content/uploads/2012/06/2012-07MassaroFinal2.pdf
Massaro, D. W. (2012). Speech Perception and Reading: Two Parallel Modes of Understanding Language and Implications for Acquiring Literacy Naturally. American Journal Psychology, 125, 307-320.
Massaro, D. W. (2012). Method And System For Representing Capitalization Of Letters While Preserving Their Category Similarity To Lowercase Letters. Patent Application Number 13/669,522, November 6. 2012.
Host: Ari Shapiro
Location: Institute For Creative Technologies (ICT) - Theatre
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Prateek Mittal (UC Berkeley): Trustworthy Communications Using Network Science
Tue, Mar 26, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Prateek Mittal, UC Berkeley
Talk Title: Trustworthy Communications Using Network Science
Series: CS Colloquium
Abstract: Our online communications are plagued by increasing threats to security and privacy. Sophisticated surveillance technologies can compromise user privacy, and the insecurity of network protocols threatens the safety of our critical infrastructure. In this talk, I argue that network science can play an important role in cybersecurity by illustrating how understanding and manipulating structural properties of networks can inform the design of trustworthy communication systems.
First, I will discuss how network structure can be leveraged to detect and isolate malicious (Sybil) accounts in online social networks. The SybilInfer system that I developed uses this approach by exploiting differences in mixing properties between benign accounts and malicious accounts. SybilInfer demonstrates how graph theoretic machine learning techniques can be applied to security problems. Second, I will discuss how specially designed network structures can help protect users' privacy by enabling them to communicate anonymously. The ShadowWalker system that I developed for anonymous communication is built around a novel network topology, which is both fast mixing and inherently verifiable. This allows ShadowWalker to scale to millions of users while being resilient to attacks on user privacy. Finally, I will conclude by highlighting the potential of leveraging complex network structures in a broad range of security and privacy problems.
Biography: Prateek Mittal is a postdoctoral scholar in Electrical Engineering and Computer Sciences at the University of California, Berkeley. His research focuses on building secure and privacy-preserving systems, drawing on techniques from applied cryptography, distributed systems, large scale machine learning and network science. His work has influenced the design of widely-used systems such as the Tor network.
He received the M.E. Van Valkenburg graduate research award for outstanding doctoral research, the Rambus Computer Engineering fellowship, and the ACM CCS 2008 outstanding paper award. He earned his M.S. and Ph.D. in Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign.
Host: Ramesh Govindan
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Manuel Egele (Carnegie Mellon University): Opposites Attract -- Static analysis on mobile apps for security and privacy
Thu, Mar 28, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Manuel Egele, Carnegie Mellon University (CyLab)
Talk Title: Opposites Attract -- Static analysis on mobile apps for security and privacy
Series: CS Colloquium
Abstract: Mobile devices are ubiquitous. Apple sold more than 400 million iOS devices to date, and it has been reported that more than 500 million Android-based devices are in customers' hands. These devices open exciting new avenues of innovation such as location-based services and mobile payment. Of course, the user has a legitimate desire to keep the privacy-sensitive data maintained and collected by these smart devices safe and secure. Unfortunately, mobile devices frequently expose such information to prying third-party applications (apps). In this talk, I will demonstrate how novel static analysis techniques can be used to automatically assess whether apps adhere to the user's expectation of privacy. My binary static analysis platform (PiOS) has the capability to evaluate many different security properties on iOS applications. For example, PiOS automatically detected numerous popular applications that leak privacy sensitive data, such as address book contents or location information over the Internet. Furthermore, based on PiOS, we were also able to retrofit iOS applications with control flow integrity protection. Android recently surpassed Apple as the most popular smart phone operating system. Thus, in this talk, I will also cover my research to leverage static analysis techniques to detect misuse of cryptographic primitives in Android apps. Furthermore, I will illustrate how these techniques can be used to refine and improve the existing coarse-grained Android permission system.
Biography: Manuel Egele is a post-doctoral researcher at Carnegie Mellon University, Cylab. Before starting at CMU, he was a post-doctoral researcher at the Computer Security Group of the Department of Computer Science at the University of California, Santa Barbara. He received his MSc (2006) and Ph.D. (2011) degrees in computer science from the University of Technology in Vienna. His research interests span numerous areas of systems security -- in particular, mobile security, privacy, and malicious code analysis. His PiOS work received a distinguished paper award at the Network and Distributed Systems Security Symposium 2011. Lately, he started investigating techniques to aid developers in avoiding common pitfalls when applying cryptographic primitives in their mobile applications.
Host: Ramesh Govindan
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
