Select a calendar:
Filter February Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for February
-
CS Colloquium: Heng Yin (Syracuse University) - A Semantics-Centric Approach to Fight Android Malware
Mon, Feb 08, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Heng Yin, Syracuse University
Talk Title: A Semantics-Centric Approach to Fight Android Malware
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
The number of new Android malware instances has grown exponentially in recent years. McAfee reports that 2.47 million new mobile malware samples were collected in 2013, which represents a 197% increase over 2012. Greater and greater amounts of manual effort are required to analyze the increasing number of new malware instances. This has led to a strong interest in developing methods to automate the malware analysis process. In this talk, I will present a series of semantics-centric techniques to fight Android malware. First of all, we need a powerful analysis framework to quickly understand the inner-working of a given malware sample. To this end, we developed a virtualization-based analysis framework called DroidScope, which can seamlessly reconstruct both OS and Java level semantic views to provide a holistic view of a malware attack. Moreover, we need to automatically classify malware samples by their functionalities and behaviors and discover zero-day malware. We proposed a new semantics-based technique for malware classification, by capturing the semantics-level behavior of an app in form of ``Weighted Contextual API Dependency Graphs". Then by computing the similarity between these graphs, we can accurately and reliably detect malware variants and zero-day malware. Furthermore, we believe that malware detection can be more effective by getting end users into the loop. In particular, we developed a new technique that can automatically generate human-readable descriptions of a given app, such that any unexpected descriptions will cause suspicions and flagged by end users. To encourage wide adoption and follow-up research, these research products are available in form of source code release and/or web services.
Biography: Heng Yin is an Associate Professor in the department of Electrical Engineering and Computer Science at Syracuse University. His research interests mainly lie in computer security. In particular, he is interested in applying program analysis techniques and virtualization techniques to improve software and system security and defeat malware attacks. He earned his PhD degree in Computer Science from the College of William and Mary in July 2009. He was a main contributor in BitBlaze team at UC Berkeley before joining Syracuse University. In 2011, he received NSF Career award.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Joseph Bonneau (Stanford) - Cryptographic transparency
Tue, Feb 09, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Joseph Bonneau, Stanford
Talk Title: Cryptographic transparency
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Traditionally, cryptography aims to eliminate trusted authorities and reduce security to computational assumptions: the system is secure as long as attackers can't guess a random key or solve a hard mathematical problem. This talk will discuss an alternate approach: retain centralized authorities but use cryptography to provide transparency that they are behaving correctly. I'll present two examples: ensuring a public key server is serving keys consistently and ensuring a Bitcoin exchange controls enough funds to be solvent. Reasoning about the security of these systems require a more holistic approach, modeling user actions and economic incentives.
Biography: Joseph Bonneau is a Postdoctoral Researcher at Stanford University and a Technology Fellow at the Electronic Frontier Foundation. His research focuses on cryptography and security protocols, particularly how they interact with human and organizational behavior and economic incentives. Recently he has focused on Bitcoin and related cryptocurrencies and secure messaging tools. He is also known for his work on passwords and web authentication. He received a PhD from the University of Cambridge under the supervision of Ross Anderson and an BS/MS from Stanford under the supervision of Dan Boneh. Last year he was as a Postdoctoral Fellow at CITP, Princeton and he has previously worked at Google, Yahoo, and Cryptography Research Inc.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: George Varghese (Microsoft Research) -From EDA to NDA: Treating Networks like Hardware Circuits
Thu, Feb 11, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: George Varghese, Microsoft Research
Talk Title: From EDA to NDA: Treating Networks like Hardware Circuits
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Surveys reveal that network outages are prevalent, and outages take hours to resolve, resulting in significant lost revenue. We suggest fresh approaches based on verification and synthesis.
First, I show how to exploit physical symmetry to scale network verification for large data centers. While Emerson and Sistla showed how to exploit symmetry for model checking in 1996, they exploited symmetry on the logical Kripke structure. We factor the symmetries into symmetries on headers and symmetries on the physical topology.
I will then describe work we have done in synthesis. I will set the stage by describing a reconfigurable router architecture called RMT and an emerging language for programming routers called P4 (that promises to extend the boundaries of Software Designed Networks). I will then describe two synthesis efforts for flexible routers, one akin to register allocation (table layout) and one akin to code generation (packet transactions). I will focus especially on code generation and show that the all-or-nothing compilation required for wire-speed forwarding requires adapting standard compiler techniques.
These results suggest that concepts from Electronic Design Automation (EDA) can be leveraged to create what might be termed Network Design Automation (NDA). I end by briefly exploring this vision. This is joint work with collaborators at Edinburgh, MSR, MIT, Stanford, and University of Washington.
Biography: George Varghese received his Ph.D. in 1992 from MIT. From 1993-1999, he was a professor at Washington University, and at UCSD from 1999 to 2013. He was the Distinguished Visitor in the computer science department at Stanford University from 2010-2011. He joined Microsoft Research in 2012. His book "Network Algorithmics" was published in December 2004 by Morgan-Kaufman. In May 2004, he co-founded NetSift, which was acquired by Cisco Systems in 2005. With colleagues, he has won best paper awards at SIGCOMM (2014), ANCS (2013), OSDI (2008), PODC (1996), and the IETF Applied Networking Prize (2013). He has won lifetime awards in networking from the EE (Kobayashi Award) and CS communities (SIGCOMM) in 2014.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Haipeng Luo (Princeton) -Optimal and Adaptive Online Learning
Tue, Feb 16, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Haipeng Luo , Princeton
Talk Title: Optimal and Adaptive Online Learning
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Online learning is one of the most important and well-established learning models in machine learning. Generally speaking, the goal of online learning is to make a sequence of accurate predictions "on the fly" when interacting with the environment. Online learning has been extensively studied in recent years, and has also become of great interest to practitioners due to its applicability to large scale applications such as advertisement placement and recommendation systems.
In this talk, I will present novel, optimal and adaptive online learning algorithms for three problems. The first problem is online boosting, a theory of boosting the accuracy of any existing online learning algorithms; the second problem is on combining expert advice more efficiently and adaptively when making online predictions; the last part of the talk is about using data sketching techniques to obtain efficient online learning algorithms that make use of second order information and have robust performance against ill-conditioned data.
Biography: Haipeng Luo is currently a fifth year graduate student working with Prof. Rob Schapire at Princeton. His main research interest is in theoretical and applied machine learning, with a focus on adaptive and robust online learning and its connections to boosting, optimization, stochastic learning and game theory. He won the Wu Prize for Excellence and two best paper awards (ICML and NIPS) in 2015.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Baris Akgun (Georgia Institute of Technology) - Robots Interactively Learning and Exploring with People
Tue, Feb 16, 2016 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Baris Akgun, Georgia Institute of Technology
Talk Title: Robots Interactively Learning and Exploring with People
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Robots are destined to move beyond the "caged" factory floors towards domains where they will be interacting closely with humans. They will encounter highly varied environments, scenarios and user demands. As a result, end-users
programming robots after deployment will be an important requirement.
In this talk, I will present results of studies with non-expert people teaching robots by demonstration and algorithms developed based on the lessons learned. My main observation is these users concentrate on achieving the goal of the demonstrated skills rather than providing good quality demonstrations. I will describe a learning from demonstration approach that leverages this goal directed behavior of users and is able to continue self-improvement on these learned models after the end-user leaves, an important step toward life-long learning. I will then talk about results of an experiment with non-expert teachers on an interactive approach that incorporates all the methods and algorithms I have introduced thus far. The work presented represents one example in a larger agenda of human-centered learning from demonstration, I conclude with a discussion of grand challenges ahead.
the lecture will be available to stream HERE. Please open link in new tab for best results.
Biography: Baris Akgun is a post-doctoral fellow at the University of Texas at Austin. He received his Ph.D. in Robotics from Georgia Institute of Technology under Assoc. Prof. Dr. Andrea Thomaz where he worked on developing algorithms and interactions that enable robots to learn from non-expert teachers and to use their learned information to autonomously get better over time. He received his M.Sc. degree in Computer Engineering in 2010 from METU working on affordance learning and mirror neuron inspired learning from demonstration under Assoc. Prof. Dr. Erol Sahin. .He received his B.Sc. degree in Mechanical Engineering with an extracurricular minor in Mechatronics in 2007. His research interests lie at the intersection of human-robot interaction and machine learning for robotics. He is currently working on deploying his developed methodologies in a household setting with non-expert teachers. His research was funded by the NSF and the ONR. He was a recipient of the Scientific and Technological Research Council of Turkey (TUBITAK) scholarship for his M.Sc and the Fulbright Scholarship for his Ph.D.
Host: CS Department
More Info: URL:https://bluejeans.com/213720883
Location: Henry Salvatori Computer Science Center (SAL) - 101
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
Event Link: URL:https://bluejeans.com/213720883
-
CS Colloquium: Arjun Radhakrishna (U. Pennsylvania) - Performance-aware Repair for Concurrent Programs
Wed, Feb 17, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Arjun Radhakrishnan, U. Pennsylvania
Talk Title: Performance-aware Repair for Concurrent Programs
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
We present a recent line of work on automated synthesis of synchronization constructs for concurrent programs. Our techniques are inspired by a study of the most common types of concurrency bugs and bugs-fixes in Linux device-drivers. As opposed to classical techniques which tend to use expensive synchronization constructs, our technique attempts to use inexpensive program transformations, such as reordering independent statements, to improve the performance of generated fixes.
Our techniques are based on the observation that a large fraction of concurrency bugs are data-independent. This observations allow us to characterize and fix concurrency bugs based only on the order of execution of the statements involved. We evaluated our techniques on several real concurrency bugs that occurred in Linux device drivers, and showed that our synthesis procedure is able to produce more efficient and "programmer-like" bug-fixes.
We finish by talk with a brief note on the general theme of soft specifications, such as performance and energy consumption, in program synthesis. Specifically, we will discuss the use of quantitative specifications and their applications to resource management in embedded and cyber-physical systems.
Biography: Arjun Radhakrishna is a post-doctoral researcher at the University of Pennsylvania. Previously, he completed his PhD at the Institute of Science and Technology, Austria advised by Prof. Thomas A. Henzinger. His research focuses primarily on using programming language techniques, specifically, automated program synthesis, for rigorous systems engineering. His current research interests include the use of alternative specification mechanisms to capture subtle soft requirements on computing systems, such as program performance, energy consumption, or a program's robustness to errors. He is also interested in verification and synthesis of concurrent programs, in particular, device drivers.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Alec Jacobson (Columbia University) - Breaking Barriers between Humans and Geometry
Thu, Feb 18, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Alec Jacobson, Columbia University
Talk Title: Breaking Barriers between Humans and Geometry
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
In the field of geometry processing, I work to make sense of existing geometric data and provide interfaces to put that data to further use. Today, we find many sources of geometric data and increasingly find useful applications effecting our daily lives. Climate analysis, self-driving cars, 3d-printed prosthetics, virtual dressing rooms and video games all share the essential tasks of collecting, processing and utilizing geometric data.
Unfortunately, barriers stand between geometric data and the people who want to analyze and understand that data. Potential consumers and content creators cannot access or edit geometry because of poor human-computer interfaces. Meanwhile, some data never reaches its intended users because processing breaks down due to lack of robustness to noise.
My long-term research goal is to dismantle the barriers between humans and geometry. In this talk, I will show how I attack this problem on both fronts. I bring ideas from differential geometry and finite-element analysis to model geometric problems more intuitively and more robustly. Meanwhile, I pursue better user interfaces to reduce human effort and increase creative or scientific exploration of geometric data. I will present my work in robust geometry processing, higher-order PDEs, real-time shape articulation, and fabricating user interfaces. Each parallel branch of investigation, while self-motivating, complements the others, and together they invite exciting new directions for future research.
Biography: Alec Jacobson is a post-doctoral researcher at Columbia University working with Prof. Eitan Grinspun. He received a PhD in Computer Science from ETH Zurich, and an MA and BA in Computer Science and Mathematics from the Courant Institute of Mathematical Sciences, New York University. His thesis on real-time deformation techniques for 2D and 3D shapes was awarded the ETH Medal and the Eurographics Best PhD award. Leveraging ideas from differential geometry and finite-element analysis, his work in geometry processing improves exposure of geometric quantities, while his novel user interfaces reduce human effort and increase exploration. He has published several papers in the proceedings of SIGGRAPH. He leads development of the geometry processing library, libigl, winner of the 2015 SGP software award.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Animashree Anandkumar (UC Irvine) - Guaranteed Non-convex Algorithms for Modern Machine Learning through Tensor Factorization
Thu, Feb 18, 2016 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Animashree Anandkumar, UC Irvine
Talk Title: Guaranteed Non-convex Algorithms for Modern Machine Learning through Tensor Factorization
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Modern machine learning involves processing massive datasets of diverse varieties such as text, images, videos, biological data, and so on. Designing efficient algorithms which are guaranteed to learn in a fast and a scalable manner is one of grand challenges. Most machine learning tasks can be cast as optimization problems, but unfortunately a majority of them are NP-hard non-convex problems. I will provide broad guidelines for overcoming this hardness barrier by: (i) focusing on conditions which make learning tractable, (ii) replacing the given optimization objective with better behaved ones, and (iii) exploiting non-obvious connections that abound in learning problems.
I will demonstrate the above guidelines using concrete examples: (i) unsupervised learning of latent variable models and (ii) training multi-layer neural networks, through a new framework involving spectral decomposition of moment matrices and tensors. Tensors are rich structures that can encode higher order relationships in data. Despite being non-convex, tensor decomposition can be solved optimally using simple iterative algorithms under mild conditions. These positive results demonstrate that previous theory on computational hardness of learning is overly pessimistic, and that we need new theoretical tools to explain the recent empirical success of non-convex learning algorithms.
This meeting will be available to stream HERE. Please right-click, open in new tab for best results.
Biography: Anima Anandkumar is a faculty at the EECS Dept. at U.C.Irvine since August 2010. Her research interests are in the areas of large-scale machine learning, non-convex optimization and high-dimensional statistics. In particular, she has been spearheading the development and analysis of tensor algorithms for a variety of learning problems. She is the recipient of several awards such as the Alfred. P. Sloan Fellowship, Microsoft Faculty Fellowship, Google research award, ARO and AFOSR Young Investigator Awards, NSF CAREER Award, Early Career Excellence in Research Award at UCI, Best Thesis Award from the ACM SIGMETRICS society, IBM Fran Allen PhD fellowship, and best paper awards from the ACM SIGMETRICS and IEEE Signal Processing societies. She received her B.Tech in Electrical Engineering from IIT Madras in 2004 and her PhD from Cornell University in 2009. She was a postdoctoral researcher at MIT from 2009 to 2010, and a visiting faculty at Microsoft Research New England in 2012 and 2014.
Host: CS Department
Webcast: https://bluejeans.com/267704433Location: Henry Salvatori Computer Science Center (SAL) - 101
WebCast Link: https://bluejeans.com/267704433
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Teamcore Seminar: Akshat Kumar (Singapore Management University) - Automated Planning and Decision Making Using Probabilistic Inference
Fri, Feb 19, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Akshat Kumar, Singapore Management University
Talk Title: Automated Planning and Decision Making Using Probabilistic Inference
Series: Teamcore Seminar
Abstract: Automated planning and decision making is a key building block of artificial intelligence. Traditionally, approaches for decision making and planning have evolved somewhat separately from that of machine learning. In this talk, I will highlight the duality between planning and learning. I will show that several decision making problems in probabilistic graphical models (such as the problem of maximum a posteriori (MAP) estimation) with applications in multiagent systems, bioinformatics, and computer vision can be viewed from a machine learning (ML) perspective using the framework of maximum likelihood estimation. Similarly, I will show how classical planning problems such as finding the shortest path in a graph can also be viewed from an ML perspective. As a result of this connection between planning and learning, I will show how probabilistic inference approaches can be used for planning, and their benefits. I will conclude by highlighting how such planning-as-inference perspective has recently helped address challenging sequential decision making problems in multiagent systems.
Host: Teamcore Group
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Seminar: Jim Boerkoel (Harvey Mudd) - Temporal Planning for Robust Human-Robot Teamwork
Mon, Feb 22, 2016 @ 12:00 AM - 01:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Jim Boerkoel, Harvey Mudd
Talk Title: Temporal Planning for Robust Human-Robot Teamwork
Series: CS Seminar Series
Abstract: Our work explores what it takes for teams of robots and humans to schedule robust interactions in the messiness of the real world.Temporal planning enables robots to automatically coordinate when the activities in their schedule should occur. In general, we want temporal plans that are adaptable to events that are beyond the direct control of agents; e.g., a robot may experience slippage or sensor failures. To do this, we must answer two questions: (1) how and when do new or unexpected events arise in practice?, and (2) how "good" is the temporal plan at adapting to unexpected events that might otherwise invalidate the plan? We have proposed a new metric called robustness, which assesses the likelihood that a multi-robot plan succeeds. We have shown that robustness is a better measure of multi-robot plan quality and that we can generate plans that optimize for robustness.
We also explore how to plan and schedule robots' interactions with humans on shared activities so that the exchanges are fluid and intuitive. For instance, one of our ongoing projects considers whether an agent can learn about the experiences of its human teammate in order to nudge them towards more optimal behavior. Another considers the roles that trust and cooperation play in human-robot interactions and how these differ from how humans trust and cooperate with each other.
Learn more: https://www.cs.hmc.edu/HEAT/
Host: Sven Koenig
Location: Henry Salvatori Computer Science Center (SAL) - 322
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Muhammad Naveed (UIUC) - Making the World a Better Place with Cryptography
Mon, Feb 22, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Muhammad Naveed, UIUC
Talk Title: Making the World a Better Place with Cryptography
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
The U.S. Department of Health and Human Services reports that the health records of up to 86% of the U.S. population have been hacked. The Ashley Madison breach revealed the private information of 37 million individuals and led to suicides and shattered families. The Apple iCloud breach led to the public release of nude photos of several celebrities. Data breaches like these abound.
In this talk, I will first describe my research toward understanding the security of existing data breach prevention systems. To thwart data breaches, property-preserving encryption has been adopted in many encrypted database systems such as CryptDB, Microsoft Cipherbase, Google Encrypted BigQuery, SAP SEEED, and the soon-to-be-shipped Microsoft SQL Always Encrypted system. To simultaneously attain practicality and functionality, property-preserving encryption schemes permit the leakage of certain information such as the relative order of encrypted messages. I will explain the practical implications of permitting such leakage, and show in real-world contexts that property-preserving encryption often does not offer strong enough security.
Next, I will describe an application-driven approach to developing practical cryptography to secure sensitive data. The approach involves collaborating with application domain experts to formulate the requirements; investigating whether a practical solution meeting the requirements is possible; and, if not, exploring the reasons behind it to relax the requirements so as to find a useful solution for the application. I will describe how I developed a cryptographic model called Controlled Functional Encryption (CFE), and how we can adopt CFE to address the privacy concerns in emerging applications such as personalized medicine.
Biography: Muhammad Naveed is a PhD candidate at UIUC studying applied cryptography and systems security. In applied cryptography, he develops practical-yet-provably-secure cryptographic systems for real applications. In systems security, he explores the fundamental security flaws in popular systems and builds defense systems. His work has had a significant impact on Android security and has helped companies such as Google, Samsung, Facebook, and Amazon secure their products and services, improving security for millions of Android users. He is the recipient of the Google PhD Fellowship in Security, the Sohaib and Sara Abbasi Fellowship, the CS@Illinois C.W. Gear Outstanding Graduate Student Award, and the best paper award at the NYU CSAW Security Research Competition. He was also a finalist in the NYU CSAW Cybersecurity Policy Competition.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Robert West (Stanford) - Human Behavior in Networks
Mon, Feb 22, 2016 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Robert West, Stanford
Talk Title: Human Behavior in Networks
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Humans as well as information are organized in networks. Interacting with these networks is part of our daily lives: we talk to friends in our social network; we find information by navigating the Web; and we form opinions by listening to others and to the media. Thus, understanding, predicting, and enhancing human behavior in networks poses important research problems for computer and data science with practical applications of high impact. In this talk I will present some of my work in this area, focusing on (1) human navigation of information networks and (2) person-to-person opinions in social networks.
Network navigation constitutes a fundamental human behavior: in order to make use of the information and resources around us, we constantly explore, disentangle, and navigate networks such as the Web. Studying navigation patterns lets us understand better how humans reason about complex networks and lets us build more human-friendly information systems. As an example, I will present an algorithm for improving website hyperlink structure by mining raw web server logs. The resulting system is being deployed on Wikipedia's full server logs at terabyte scale, producing links that are clicked 10 times as frequently as the average link added by human Wikipedia editors.
Communication and coordination through natural language is another prominent human network behavior. Studying the interplay of social network structure and language has the potential to benefit both sociolinguistics and natural language processing. Intriguing opportunities and challenges have arisen recently with the advent of online social media, which produce large amounts of both network and natural language data. As an example, I will discuss my work on person-to-person sentiment analysis in social networks, which combines the sociological theory of structural balance with techniques from natural language processing, resulting in a sentiment prediction model that clearly outperforms both text-only and network-only versions.
I will conclude the talk by sketching interesting future directions for computational approaches to studying and enhancing human behavior in networks.
The lecture will be available to stream HERE. Please Open in New Tab for best results.
Biography: Robert West is a sixth-year Ph.D. candidate in Computer Science in the Infolab at Stanford University, advised by Jure Leskovec. His research aims to understand, predict, and enhance human behavior in social and information networks by developing techniques in data science, data mining, network analysis, machine learning, and natural language processing. Previously, he obtained a Master's degree from McGill University in 2010 and a Diplom degree from Technische Universität München in 2007.
Host: CS Department
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Vincent Liu (U. Washington) - Improving the Cost and Reliability of Data Center Networks
Tue, Feb 23, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Vincent Liu , U. Washington
Talk Title: Improving the Cost and Reliability of Data Center Networks
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
In recent years, data center networks have grown to an unprecedented scale. The largest of these are expected to connect hundreds of thousands of servers and are expected to do so with high reliability and low cost. The current solution to these problems is to use an idea first proposed for telephone networks in the early 1950's: Clos network topologies. These topologies have a number of substantial benefits, but their use in this new domain raises a set of questions.
In this talk, I will present two systems that make small changes to state-of-the-art data center designs to provide large improvements to performance, reliability, and cost. I will first describe F10, a data center architecture that can provide both near-instantaneous reaction to failures and near-optimal handling of long-term load balancing. Central to this architecture is a novel network topology that provides all of the benefits of a traditional Clos topology, but also admits local reaction to and recovery from failures. I will also describe Subways, a network architecture that looks at how to use multiple network interfaces on each server to handle growth and performance issues in today's data centers.
Biography: Vincent Liu is a PhD candidate in Computer Science and Engineering at the University of Washington. Before that, he completed his undergraduate degree in Computer Science at the University of Texas at Austin. His research is in the general area of networked systems across all layers of the networking stack, from hardware concerns to application and workload modeling. He has published in a variety of fields including data center networks, fault-tolerant distributed systems, energy-efficient wireless communication, and systems to preserve security and privacy. His work has won Best Paper Awards at NSDI 2013, ACM SIGCOMM 2013, and NSDI 2015. He was also awarded a Google PhD fellowship and Qualcomm Innovation Fellowship.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Giuseppe Loianno (U. of Pennsylvania) - Flying Robots: Fast Autonomous Navigation and Physical Interaction
Tue, Feb 23, 2016 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Giuseppe Loianno, U. of Pennsylvania
Talk Title: Flying Robots: Fast Autonomous Navigation and Physical Interaction
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Flying Robots are starting to play a major role in several tasks such as search and rescue, interaction with the environment, inspection and monitoring.
Unfortunately, their dynamics make them extremely difficult to control and this is particularly true in absence of external positioning systems, such as GPS and motion-capture systems. Additionally, autonomous maneuvers based on onboard sensors are still very slow compared to those attainable with motion capture systems.
It is essential to provide fast autonomous capabilities to Micro Aerial Vehicles (MAVs) to fly in a reliable way in unstructured environments.
This is achieved giving the vehicle, the ability to identify its state, using either absolute and relative asynchronous measurements provided by different sensors. Those have to be fused, exploiting different rates and statistical sensors' properties.
Algorithms that rely on the combination of a camera and IMU, two lightweight and inexpensive sensors, represent a valid solution for autonomous navigation and environment interaction, especially in case of small scale vehicles.
In this talk, I will give an overview of my research activities toward fast autonomous navigation of MAVs, collaboration between multiple aerial vehicles and environment interaction using a camera and IMU as the main sensor modalities, showing how these problems can be even solved by the use of classic consumer electronic devices.
This lecture will be available to stream HERE. Please open in new tab for best results.
Biography: Giuseppe Loianno received B.Sc and M.Sc in Automation Engineering both with honours at University of Naples Federico II in December 2007 and February 2010, respectively. He has been, during the academic year 2008, an exchange student at KTH (Royal Insitute of Technology) in Stockholm.
He developed his master thesis at ETH Zurich at the ASL laboratory focusing on Micro Aerial Vehicles under the supervision of Prof. Dr. Davide Scaramuzza. He received his Ph.D in computer and automatic engineering focusing in Robotics in May 2014 in the PRISMA Lab group, led by Prof. Dr. Bruno Siciliano. From April 2013 he worked as vising Ph.D student for 14 months with the Grasp Lab at University of Pennsylvania, supervised by Prof. Dr. Vijay Kumar and working for 12 months a postdoctoral researcher. He is currently a research scientist. His research interests include visual navigation, sensor fusion and visual servoing for micro aerial vehicles.
Host: CS Department
Location: Henry Salvatori Computer Science Center (SAL) - 101
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Chao Wang (Virginia Tech) - Symbolic Analysis for Detecting and Mitigating Errors in Concurrent Software
Thu, Feb 25, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Chao Wang, Virginia Tech
Talk Title: Symbolic Analysis for Detecting and Mitigating Errors in Concurrent Software
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
The use of multi-core architecture is now pervasive, spanning from embedded systems and smart phones, to commodity PCs, all the way to high-end servers and distributed systems. As such, developers must write concurrent software. However, writing correct and efficient concurrent software is difficult. Although automated analysis to aid in their development would be invaluable, existing methods are either fast but inaccurate, or accurate but slow, due to the inherent difficulty in circumventing the path and interleaving explosion problem. In this talk, I will introduce a series of symbolic predictive analysis methods for analyzing concurrent software. This analysis consists of two steps. First, we derive a predictive model from the execution traces collected at run time as well as the software code. The model captures not only the given executions but also the alterative interleavings of events in these executions. Then, we use symbolic analysis to check if errors exist in these alternative interleavings. This is accomplished by capturing these interleavings and the error conditions using a set of logic formulas and deciding them using a Satisfiability Modulo Theory (SMT) solver. Although our primary focus is to reduce the cost associated with analyzing and verifying concurrent software, the predictive model and related analysis techniques are also useful in addressing issues related to performance and security.
Biography: Chao Wang is an Assistant Professor of the ECE Department and the CS Department (by courtesy) of Virginia Tech. He received the ONR Young Investigator award in 2013 and the NSF CAREER award in 2012. His area of specialization is Software Engineering and Formal Methods, with emphasis on concurrency, formal verification, and program synthesis. He published a book and more than fifty papers in top venues of related field including ICSE, FSE, ASE, ISSTA, CAV, PLDI, and POPL. He received the FMCAD Best Paper award in 2013, the ACM SIGSOFT Distinguished Paper award in 2010, the ACM TODAES Best Paper of the Year award in 2008, and the ACM SIGDA Outstanding PhD Dissertation award in 2004. Dr. Wang received his PhD degree from the University of Colorado at Boulder in 2004. From 2004 to 2011, he was a Research Staff Member at NEC Laboratories of America in Princeton, NJ, where he received a Technology Commercialization award in 2006.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
ICT Distinguished Lecture Series
Thu, Feb 25, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science, USC Viterbi School of Engineering
Conferences, Lectures, & Seminars
Speaker: James W. Pennebaker, Social Psychologist and the Centennial Liberal Arts Professor of Psychology at the University of Texas at Austin
Talk Title: The synchronous massive online course (SMOC) and new model of online education
Abstract: Growing out of research on social and personality psychology and computerized text analysis, Pennebaker and his colleague Sam Gosling have developed a live online class that relies on daily testing, small online discussion groups, and a TV talk show format. Based on four classes ranging from 800 to 1500 students, significant gains were seen in performance in the class over previous courses. More striking, students taking the SMOC do better in courses in subsequent semesters and achieve large reductions in the achievement gap.
Biography: James Pennebaker is the Regents Centennial Professor of Psychology and Executive Director of Project 2021 at the University of Texas at Austin. Author of over 300 publications and 9 books, his work on physical symptoms, expressive writing, and language psychology is among the most cited in psychology and the social sciences. He has received multiple research and teaching awards.
Light refreshments will be served.
Host: Stefan Scherer, ICT
Location: Institute For Creative Technologies (ICT) - Theater
Audiences: Everyone Is Invited
Contact: Orli Belman/ICT
-
CS Colloquium: Hyun Oh Song (Stanford) -Beyond supervised pattern recognition: Efficient learning with latent combinatorial structure
Thu, Feb 25, 2016 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Hyun Oh Song, Stanford
Talk Title: Beyond supervised pattern recognition: Efficient learning with latent combinatorial structure
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Supervised pattern recognition with 10^6 training data and 10^9 layered parameters has brought tremendous advances in artificial intelligence. However, there are two main limitations to this approach: 1) The knowledge learned in one area doesn't easily transfer to another and 2) supervising every single task is not only infeasible but also requires huge amounts of human labeled data which is costly and time consuming. In this talk, I will suggest a unifying framework which jointly reasons the prediction variable and the underlying latent combinatorial structure of the problem as a way to address such limitations. To demonstrate the practical benefits of the approach, we explore classification, localization, clustering, and retrieval tasks under settings that go beyond fully supervised pattern recognition.
Biography: Hyun Oh Song is a postdoc in SAIL in the computer science department at Stanford University. He received Ph.D. in Computer Science at UC Berkeley in late 2014 under the supervision of Prof. Trevor Darrell. He is a recipient of five year Ph.D. fellowship from Samsung Lee Kun Hee Scholarship Foundation. His research interest lies at the intersection between machine learning, computer vision, and optimization. He has an academic website at http://ai.stanford.edu/~hsong.
Host: CS Department
Webcast: https://bluejeans.com/501895444Location: Henry Salvatori Computer Science Center (SAL) - 101
WebCast Link: https://bluejeans.com/501895444
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Seminar: Jim Hansen (Naval Research Laboratories) - Illicit Trafficking Surveillance and Interdiction: INTEL and METOC-informed Decision Guidance
Fri, Feb 26, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Jim Hansen, Naval Research Laboratories
Talk Title: Illicit Trafficking Surveillance and Interdiction: INTEL and METOC-informed Decision Guidance
Series: CS Seminar Series
Abstract: Many Navy problems can be cast as a variant of "I have some assets with various capabilities and constraints that I need to distribute tomorrow in order to best carry out my mission. Which of them should go where and do what?" Factors that influence the decision include available intelligence information, an understanding of the adversary's expected behavior, and the environmental conditions. This talk will describe a general approach that has been applied to the pirate problem, illicit (drug) trafficking, and anti-submarine warfare. It will also include a discussion of challenges associated with working with users and transitioning research projects to operations.
Host: Computer Science Department
Location: Ronald Tutor Hall of Engineering (RTH) - 105
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Sameer Singh (University of Washington) - Interactive Machine Learning for Information Extraction
Mon, Feb 29, 2016 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Sameer Singh, University of Washington
Talk Title: Interactive Machine Learning for Information Extraction
Series: CS Colloquium
Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium
Most of the world's knowledge, be it factual news, scholarly research, social communication, subjective opinions, or even fictional content, is now easily accessible as digitized text. Unfortunately, due to the unstructured nature of text, much of the useful content in these documents is hidden. The goal of "information extraction" is to address this problem: extracting meaningful, structured knowledge (such as graphs and databases) from text collections. The biggest challenges when using machine learning for information extraction include the high cost of obtaining annotated data and lack of guidance on how to fix mistakes.
In this talk, I propose interpretable representations that allow users and machine learning models to interact with each other: enabling users to inject domain knowledge into machine learning, and providing explanations for why a machine learning model is making a specific prediction. I study these techniques using relation extraction as the application, an important subtask of information extraction where the goal is to identify the types of relations between entities that are expressed in text. I first describe how symbolic domain knowledge, if provided by the user as first-order logic statements, can be injected into relational embeddings to improve the predictions. In the second part of the talk, I present an approach to "explain" any machine learning prediction using a symbolic representation, which the user may annotate directly for more effective supervision. I present experiments that demonstrate that an interactive interface between a user and machine learning is effective in reducing annotation effort and in quickly training accurate extraction systems.
Biography: Sameer Singh is a Postdoctoral Research Associate at the University of Washington, working on large-scale and interactive machine learning applied to information extraction and natural language processing. He received his PhD from the University of Massachusetts, Amherst, during which he also interned at Microsoft Research, Google Research, and Yahoo! Labs on massive-scale machine learning. He was recently selected as a DARPA Riser, won the grand prize in the Yelp dataset challenge, has been awarded the Yahoo! Key Scientific Challenges and the UMass Graduate School fellowships, and was a finalist for the Facebook PhD fellowship.
Host: CS Department
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
