Type of Talk:  UTCS Distinguished Lecture Series

Speaker/Affiliation:  Richard Stallman/Free Software Foundation (gnu.org)

Date/Time:  Friday, April 24, 2009  2:15 p.m. - 4:15 p.m.

Location:  WEL 1.308

Host:  J Strother Moore

Talk Title:  "Copyright vs Community in the Age of Computer Networks"

Talk Abstract:

Copyright developed in the age of the printing press, and was designed to fit with the system of centralized copying imposed by the printing press. But the copyright system does not fit well with computer networks, and only draconian punishments can enforce it.

The global corporations that profit from copyright are lobbying for draconian punishments, and to increase their copyright powers, while suppressing public access to technology. But if we seriously hope to serve the only legitimate purpose of copyright--to promote progress, for the benefit of the public--then we must make changes in the other direction.

Speaker Bio:

Richard Stallman launched the development of the GNU operating system (see www.gnu.org) in 1984. GNU is free software: everyone has the freedom to copy it and redistribute it, as well as to make changes either large or small. The GNU/Linux system, basically the GNU operating system with Linux added, is used on tens of millions of computers today. Stallman has received the ACM Grace Hopper Award, a MacArthur Foundation fellowship, the Electronic Frontier Foundation's Pioneer award, and the Takeda Award for Social/Economic Betterment, as well as several honorary doctorates.

There is a sign up schedule for this event:

http://www.cs.utexas.edu/department/webevent/utcs/events/cgi/eidshow.cgi?person=DaphneKoller

Type of Talk:  UTCS Distinguished Lecturer

Speaker/Affiliation:  Daphne Koller/Stanford University

Date/Time:  Tuesday, April 14, 2009 11:00 a.m.

Location:  ACES 2.302

Host: J. Strother Moore

Talk Title:  "Probabilistic Models for Holistic Scene Understanding"

Talk Abstract:

Over recent years, computer vision has made great strides towards annotating parts of an image with symbolic labels, such as object categories (things) or segment types (stuff). However, we are still far from the goal of providing a semantic description of an image, such as "a man, walking a dog on a sidewalk, carrying a backpack".  In this talk, I will describe some projects we have done that attempt to use probabilistic models to move us closer towards the goal.

The first part of the talk will present methods that use a more holistic  scene analysis to improve our performance at core tasks such as object detection, segmentation, or 3D reconstruction.   The second part of the talk will focus on finer-grained modeling of object shape, so as to allow us to annotate images with descriptive labels related to the object shape, pose, or activity (e.g., is a cheetah running or standing).   These vision tasks rely on novel algorithms for core problems in machine learning and probabilistic models, such as
efficient algorithms for probabilistic correspondence, transfer learning across related object classes for learning from sparse data, and more.

Speaker Bio:
Daphne Koller is a Professor of Computer Science at Stanford University.  Her main research focus is in developing and using machine learning and probabilistic methods to model and analyze complex systems, and she is particularly interested in using these techniques to understand biological systems and the world around us. Professor Koller is the author of over 100 refereed publications, which have appeared in venues that include Science, Nature Genetics, and the Journal of Games and Economic Behavior.  She is a Fellow of
the American Association for Artificial Intelligence, and has received a number of awards, including the Sloan Foundation Faculty Fellowship in 1996, the ONR Young Investigator Award in 1998, the Presidential Early Career Award for Scientists and Engineers (PECASE) from President Clinton in 1999, the IJCAI Computers and Thought Award in 2001, the Cox Medal for excellence in fostering undergraduate research at Stanford in 2003,  the MacArthur Foundation Fellowship in 2004 and the first-ever ACM/Infosys award in 2008.

There is a sign up schedule for this talk: http://www.cs.utexas.edu/department/webevent/utcs/events/cgi/eidshow.cgi?person=JonHowell

Type of Talk:  UTCS Colloquium

Speaker/Affiliation:  Jon Howell/Microsoft Research

Date/Time:  Monday, April 20, 2009  3:30 p.m.

Location:  ACES 2.402

Host: Mike Walfish

Talk Title: "Leveraging Legacy Code to Deploy Desktop Applications on the Web"

Talk Abstract:

Xax is a browser plugin model that enables developers to leverage existing tools, libraries, and entire programs to deliver feature-rich applications on the web. Xax employs a novel combination of mechanisms that collectively provide security, OS-independence, performance, and support for legacy code. These mechanisms include memory-isolated native code execution behind a narrow syscall interface, an abstraction layer that provides a consistent binary interface across operating systems, system services via hooks to existing browser mechanisms, and lightweight modifications to existing tool chains and code bases. We demonstrate a variety of applications and libraries from existing code bases, in several languages, produced with various tool chains, running in multiple browsers on multiple operating systems. With roughly two person-weeks of effort, we ported 3.3 million lines of code to Xax, including a PDF viewer, a Python interpreter, a speech synthesizer, and an OpenGL pipeline.

PDF: http://research.microsoft.com/pubs/72878/xax-osdi08.pdf 

Speaker Bio:

Jon works at the intersection of security and scalability in distributed systems.  His recent projects focus on the convergence of utility computing and web-delivered applications.  He has worked on the novel Asirra Captcha, a rigorous proof of correctness for Paxos, the Farsite distributed filesystem, and cross-domain authentication.

Type of Talk:  UTCS Bruce Porter Sponsored

Speaker/Affiliation: Steve Hayman/Apple

Date/Time:  Tuesday, March 31, 2009 9:00 a.m. - 11:00 a.m.

Location:  ACES 2.302

Host:  Bruce Porter

Talk Title:  "iPhone Under the Hood"

Talk Abstract:

Developing for the iPhone with Apple's SDK

Please join Apple for an introduction to developing for the hottest mobile platform on the planet.

Steve Hayman, Apple Consulting Engineer and developer tools expert, will provide a fact-filled and entertaining presentation on the tools and programs that are available for developing applications on the iPhone and iPod touch.  Steve will explain the various developer programs and demonstrate the tools that you can use to quickly develop applications for your institution and for the Apple App store.

If you are considering adding the iPhone development to your curriculum, building applications for your school or are curious as to what all the fuss is about, please plan on attending this unique presentation

There is a sign up schedule for this event

http://www.cs.utexas.edu/department/webevent/utcs/events/cgi/eidshow.cgi?person=PradeepRavikumar-FACULTYCANDIDATE

Type of Talk:  UTCS FACULTY CANDIDATE

Speaker/Affiliation:   Pradeep Ravikumar/University of California-Berkeley

Date/Time:  Tuesday, March 31, 2009  11:00 a.m.

Location:  ACES 2.302

Host:  Inderjit Dhillon

Talk Title:  "Sparse Model Estimation: Parametric and Nonparametric Settings"

Talk Abstract:

 A common approach in settings with high-dimensional data has been to estimate models that are "sparse," in the sense that an index set of relevant model components has small cardinality. In this talk I will cover two instances, one parametric and the other nonparametric, of sparse model estimation.

The first part of the talk considers the task of estimating the covariance and inverse covariance or concentration matrices of a random vector from i.i.d. observations. We study an estimator based on minimizing an l1-penalized log-determinant Bregman divergence, that is equivalent to the usual l1-regularized maximum likelihood estimator when the random vector is multivariate Gaussian. We analyze the performance of this estimator under high-dimensional scaling, in which the number of variables and other model parameters are allowed to grow as a function of the sample size.  Our analysis identifies key players affecting the convergence rates of the estimator in various norms as well as its success in recovering the true sparsity pattern (its "sparsistency").

The second part of the talk considers the task of encoding fMRI signals from the primary visual cortex, also called area V1, of the brain in response to natural image stimuli; as well as identifying potential features of images that drive the neural activity. Our method is based on the understanding that the fMRI signal reflects the pooled, and potentially nonlinearly transformed output of a large population of neurons in area V1. Our class of models, which we call the V-SPAM framework, mimics this with an initial hierarchical filtering stage that consists of three layers of artificial neuronal cells, and a final nonparametric pooling stage which learns nonparametric transformations of a sparse set of neuronal filters.

This is joint work with Garvesh Raskutti, Vincent Vu, Martin Wainwright, Bin Yu, and the Jack Gallant lab at UC Berkeley; Kendrick Kay, Thomas Naselaris and Jack Gallant.

There is a <a target="UTCSsignupSchedule" href="http://www.cs.utexas.edu/department/webevent/utcs/events/cgi/eidshow.cgi?person=JamesBalfour-CSandECEFACULTYCANDIDATE">signup schedule</a> for this event (UT EID required).

Type of Talk:  UTCS & ECE FACULTY CANDIDATE

Speaker/Affiliation:  James Balfour/Stanford University

Date/Time:  Monday, April 6, 2009  2:00 p.m.

Location:  ACES 2.302

Host:  Steve Keckler and Yale Patt

Talk Title:  "ELM - Parallel Architectures for Efficient Embedded Computing"

Talk Abstract:

Embedded systems are ubiquitous, appearing in applications as diverse as mobile phones, digital television, automobiles, communications systems, sensor networks, and medical devices.  Satisfying the demanding computational and efficiency requirements of such systems presently requires the use of complex application-specific integrated circuits (ASICs). Designing ASICs is time consuming and expensive, often involving tasks such as lowering reference implementations from C to equivalent gate-level representations. The significant design and verification efforts required impose large non-recurring engineering costs in the development of new systems, deterring innovation and limiting the viability of new applications.


Programmable processors offer flexibility, improved productivity, and  reduced cost. Unfortunately, even low-power embedded processors are  inefficient compared to ASICs.  The inefficiency arises because  programmable processors devote significantly more energy and chip area  to delivering instructions and data to functional units.  To become viable ASIC replacements, programmable processors must deliver high  computational performance with significantly greater efficiency.


In this talk, I will describe ELM, an energy-efficient programmable  processor for high-performance embedded applications.  ELM is 10x more  efficient that conventional embedded processors, and approaches the  efficiency of ASICs on compute-intensive tasks.  ELM achieves this  efficiency by using distributed and hierarchical register and memory  organizations that allow software to better exploit parallelism and  instruction/data locality in applications.  This talk will focus on  novel computer architecture, compiler, and circuit techniques used in  ELM, and discuss how the systems-based approach used to design ELM can  be used to improve the efficiency of embedded systems more broadly.
 

There is a sign up schedule for this event:

http://www.cs.utexas.edu/department/webevent/utcs/events/cgi/eidshow.cgi?person=AllenYang-FACULTYCANDIDATE

Type of Talk:  UTCS FACULTY CANDIDATE

Speaker/Affiliation:  Allen Yang/University of California-Berkeley

Date/Time:  Thursday, April 9, 2009  11:00 a.m.

Location:  ACES 2.302

Host:  Kristen Grauman

Talk Title: "High-Dimensional Multi-Model Estimation -- Its Algebra, Statistics, and Sparse Representation"

Talk Abstract:

 Recent advances in information technologies have led to unprecedented large amounts of high-dimensional data from many emerging applications. The need for more advanced techniques to analyze such complex data calls for shifting research paradigms. In this talk, I will overview and highlight several results in the area of mixture-model estimation in high-dimensional data spaces. Applications will be presented such as motion segmentation, image segmentation, face recognition, and human action categorization. Through this talk, I intend to emphasize the confluence of algebra and statistics that may lead to more advanced solutions in analyzing complex, singular data structures such as mixture linear subspaces and nonlinear manifolds.

In the first part of the talk, I will introduce an algebro-geometric technique to simultaneously segment mixture linear and nonlinear manifolds, and applications in image-based motion segmentation and texture segmentation. The solutions are robust the moderate data noise and outliers, and outperform classical model-estimation methods such as RANSAC and Normalized-Cut. The second part will be focused on classification of mixture subspace models and its application in face recognition, where the prior information about the subspaces is provided through training examples. Inspired by compressive sensing theory, the recognition problem is reformulated via a sparse representation. Furthermore, efficient solutions exist to recover such sparse representation under high data distortion using L-1 minimization. Finally, I will discuss several open problems in the emerging field of distributed sensor perception.
 

Type of Talk:  UTCS Colloquium/Workshop

Speaker/Affiliation:   2009 Workshop on Distributed Computing/PODC 2009 Program Committee

Date/Time:  Friday, April 3, 2009 8:30 a.m. - 5:00 p.m.

Location:  ACES 2.402

Host:  Lorenzo Alvisi

Talk Title:  "Work-in-Progress Talks by Distributed Computing Researchers"

Talk Abstract:

The ACM SIGACT-SIGOPS Symposium on Principles of Distributed Computing (PODC) focuses on research in the theory, design, specification, and implementation of distributed systems.

The occasion of the PODC 2009 program committee meeting at UT-Austin on April 2, 2009 brings together top academic and industrial researchers in distributed computing.

The 2009 Workshop on Distributed Computing, Friday, April 3, 2009, presents work-in-progress talks by several PODC program committee members. The Workshop is an exciting opportunity to sample the newest research by the top practitioners in the field, including
Ittai Abraham (MSR Silicon Valley, USA)
Marcos Aguilera (MSR Silicon Valley, USA)
Hagit Attiya (Technion, Israel)
Rida Bazzi (Arizona State U, USA)
Christian Cachin (IBM Zurich, Switzerland)
Rachid Guerraoui (EPFL, Switzerland)
Tim Harris (MSR Cambridge, UK)
Anne-Marie Kermarrec (INRIA, France)
Dahlia Malkhi (MSR Silicon Valley, USA)
Alessandro Panconesi (Sapienza U, Italy)
Sergio Rajsbaum (UNAM, Mexico)
Andrea Richa (Arizona State U, USA)
Tim  Roughgarden (Stanford, USA)
Marko Vukolic (IBM Zurich, Switzerland)
Haifeng Yu (U Singapore, Singapore)

A schedule is available at
http://www.cs.utexas.edu/~lasr/09PODCpc/workshop.html

The workshop is free and open to the public; we welcome your attendance.  No registration is necessary. The workshop will be held in the Applied Computational Engineering and Sciences Building (ACES), Rm 2.402 on the UT campus at the southeast corner of Speedway and 24th St.

Directions: http://www.cs.utexas.edu/~lasr/home/contact.html

The workshop is hosted by Professor Lorenzo Alvisi of the Lab for Advanced Systems Research (LASR) in the UT-Austin Dept of Computer Sciences. We gratefully acknowledge Indeed <www.indeed.com> for their sponsorship of the 2009 Workshop on Distributed Computing.

Type of Talk:  UTCS Colloquium

Speaker/Affiliation:  Devi Parikh/Carnegie Mellon University

Location: ACES 3.408

Host:  Kristen Grauman

Talk Title:  "The Role of Context in Image Understanding: When, For What, and How?"

Talk Abstract:

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Colloquium

Speaker/Affiliation:  Hao Jiang/Boston College

Date/Time:  Wednesday, April 8, 2009 11:00 a.m.

Location:  ACES 2.404B

Host:  Kristen Grauman

Talk Title: "Labeling Problems and Convex Solutions"

Talk Abstract:

Many problems in computer vision such as image matching and object tracking can be abstracted into labeling problems in which we try to
assign labels to model parts in some consistent fashion. Apart from special cases which can be solved efficiently using polynomial algorithms,  general consistent labeling problems are NP-hard. We are interested in efficient approximation algorithms that can explore the search space    globally and approach the global optimum. Convex approximations are useful methods that meet this demand. By carefully constructing convex approximations, we can achieve efficient solutions which in some cases are even faster than greedy methods. I will show how this approach helps    in solving the problems of scale and rotation invariant object matching, action recognition, human pose estimation and multiple object tracking.

Speaker Bio:
Hao Jiang received PhD from Simon Fraser University in 2006. He was an postdoctoral research fellow in the University of British Columbia in
2006-2007 and an associate researcher in Microsoft Research Asia in 1999-2000. He is currently an assistant professor in the computer science department at Boston College. His research interests include computer vision, artificial intelligence, multimedia and graphics.
 

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Colloquium/AI

Speaker/Affiliation:  Joohyung Lee/Arizona State University

Date/Time: Friday, April 3, 2009 11:00 a.m.

Location:  TAY 3.144

Host:  Vladimir Lifschitz

Talk Title:  "Circumscriptive Event Calculus as Answer Set Programming"

Type of Talk:  UTCS Colloquium

Speaker/Affiliation:   Susan Hohenberger/John Hopkins University

Date/Time:  Friday, April 17, 2009  11:00

Location:  TAY 3.128

Host:  David Zuckerman

Talk Title:  "Short Signatures from Standard Assumptions"

Talk Abstract:

Digital signatures are fundamental for modern authentication.  Currently, almost all ``short'' signatures rely on the random oracle heuristic or depend upon strong (and relatively new) complexity assumptions.   In this talk, we present new short signature schemes in the standard model (i.e., without random oracles) based upon the traditional RSA and Computational Diffie-Hellman assumptions.

Our construction method takes two different approaches.   First, we show how to realize these signatures in a stateful setting, where the signer must store and associate each signature with an index that represents how many signatures that signer has issued up to that point.   In this setting, we realize two new short signature schemes under the RSA assumption and Computational Diffie-Hellman assumption in bilinear groups.

Next, we remove the need for the signer to keep state, by developing a new proof technique that allows the simulator to predict a prefix of the message on which the adversary will forge, and then use this knowledge to embed the challenge.   In this setting, we develop a signature secure under the RSA assumption, which requires only one element of Zn* and one integer.  We also provide an entirely new security analysis for the Waters signatures, which are secure under the CDH assumption in bilinear groups.

Speaker Bio:

Susan Hohenberger is an Assistant Professor at Johns Hopkins University.    She received a Ph.D. from MIT in 2006, where she was advised by Ronald Rivest, and a B.S. from The Ohio State University in 2000.   Prior to joining Hopkins, she completed a post-doc at IBM Zurich Research.   Susan's primary research interests are in cryptography and computer security.   She is a recipient of a 2008 Microsoft Research New Faculty Fellowship.

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Colloquium

Speaker/Affiliation:  Cordelia Shmid/INRIA

Date/Time:  Monday, April 27, 2009 11:00 a.m.

Location:  TAY 3.128

Host:  Kristen Grauman

Talk Title:  "Learning Visual Human Actions from Movies"

Talk Abstract:

We address the problem of recognizing natural human actions in diverse and realistic video settings. This challenging but important subject has mostly been ignored in the past due to several problems one of which is the lack of realistic and annotated video datasets. Our first contribution is to address this limitation and to investigate the use of movie scripts for automatic annotation of human actions in videos. We evaluate alternative methods for action retrieval from scripts and show benefits of a text-based classifier. Using the retrieved action samples for visual learning, we next turn to the problem of action classification in video. We present a new method for video classification that builds upon and extends several recent ideas including local space-time features, space-time pyramids and multi-channel non-linear SVMs. Furthermore, we show how to mine relevant context information and use it to improve action recognition. We finally apply the method to learning and classifying challenging action classes in movies and show promising results.

This is joint work with I. Laptev, M. Marszalek and B. Rozenfeld.
 

Type of Talk:  UTCS & ECE Corporate Connection

Speaker/Affiliation:  Dr. Raghu Ramakrishnan/Yahoo!

Date/Time:  Wednesday, April 22, 2009 11:00 a.m.

Location:  ACES 2.302

Host:  UTCS & ECE

Talk Title:  "Semantics on the Web: How do we get there?

Talk Abstract:

It is becoming increasingly clear that the next generation of web search and advertising will rely on a deeper understanding of user intent and task modeling, and a correspondingly richer interpretation of content on the web. How we get there, in particular, how we understand web content in richer terms than bags of words and links, is a wide open and fascinating question. I will discuss some of the options here, and look closely at the role that information extraction can play. 

Speaker Bio:

Dr. Raghu Ramakrishnan is Chief Scientist for Audience and Cloud Computing at Yahoo!, and is a Research Fellow, heading the Web Information Management group in Yahoo! Research. His work in database systems, with a focus on data mining, query optimization, and web-scale data management has influenced query optimization in commercial database systems and the design of window functions in SQL:1999. His paper on the Birch clustering algorithm received the SIGMOD 10-Year Test-of-Time award, and he has written the widely-used text "Database Management Systems" (with Johannes Gehrke). Ramakrishnan is a Fellow of the ACM and IEEE and has received the ACM SIGKDD Innovations Award, the ACM SIGMOD Contributions Award, a Distinguished Alumnus Award from IIT Madras, a Packard Foundation Fellowship, and an NSF Presidential Young Investigator Award.

He is Chair of ACM SIGMOD, on the Board of Directors of ACM SIGKDD and the Board of Trustees of the VLDB Endowment, and is on leave as Professor of Computer Sciences at the University of Wisconsin-Madison. He was founder and CTO of QUIQ, a company that pioneered question-answering communities, powering Ask Jeeves' AnswerPoint as well as customer-support for companies such as Compaq.
 

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Distinguished Speaker Lecture Series

Speaker/Affiliation:  Madhu Sudan/MIT

Date/Time:  Tuesday, April 28, 2009 11:00 a.m.

Location:  ACES 2.302

Host:  David Zuckerman

Talk Title:  "Communication and Computation: A Need for a New Unifying Theory?"

Talk Abstract:

The theories of computing (Turing, ~1930s) and communication (Shannon, Hamming ~1940s) have had a profound impact on the development of the two fields and the resulting technologies have drastically altered our lives today.  Part of the success of the two theories can be attributed to a clean separation of the computing elements from the communicating elements.  Today, however, communication and computing are coming ever closer together, often leaving the human out of the loop.  This merger is posing new challenges, definitional and algorithmic, to the theory of communication.  In this talk I will describe some of the concrete issues that our group has focussed on, and our attempts at modelling these problems and, in some cases, describe some preliminary solutions.

Speaker Bio:

Type of Talk:  UTCS Colloquium and ACT Seminar

Speaker/Affiliation:  Madhu Sudan/ MIT

Date/Time:  Friday, May 1, 2009  11:00 a.m.

Location:  ACES 2.402

Host:  David Zuckerman

Talk Title:  "The Role of Invariance in Property Testing"

Talk Abstract:

Property testing considers the task of testing if some given data satisfies a desired property by sampling the data probabilistically in very few places. The ``oldest'' property test might be the use of polling to predict the outcome of an upcoming election. Modern research has extended the scope of property tests to a much richer class of properties including tests of linearity ("is the data essentially linear with respect to some parameters"), multilinearity, low-degreeness, colorability ("is the data describing a graph with small chromatic number") etc.

What makes some properties testable so efficiently, that we do not have to look at the entire data in order to test for it? We suggest that for interesting properties, testability ought to be related to the "invariances" shown by the property: i.e., if the data is viewed as a function from some input to some output, then the "invariances" are given by a set (a group) of permutations of the input space under which the property is invariant. We then investigate this hypothesis in the context of "algebraic properties".

The class of algebraic properties we consider are linear properties (if f, g satisfy the property, then so does f+g) mapping a vector space K^n over a field K to a subfield F.  The invariance we study is that of "linear-invariance" (the property is invariant under linear transformations of the domain).  Linear invariant properties lead to broad extensions of the algebraic properties of linearity and low-degreeness. We show that when $K= O(1)$, linear-invariant properties are testable with constant queries if and only if they possess O(1)-local characterizations.  This leads to a unifying proof of previous algebraic property tests (though not in their most "efficient" forms) while leading to many new properties that were not known to be testable before. In the talk, we will attempt to highlight some of the interesting aspects of linear-invariant properties, and describe our new, unifying, analysis.

Based on joint works with Elena Grigorescu and Tali Kaufman (MIT).

Speaker Bio:

Madhu  Sudan  received  his  Bachelor's  degree  from  the Indian Institute of Technology at New Delhi in 1987 and his  Ph.D. from the University of California at Berkeley in 1992. From 1992-1997 he was a Research Staff Member at IBM's Thomas J. Watson Research Center. In 1997, he moved to MIT where he is now the Fujitsu Professor of Electrical Engineering and Computer Science, and Associate Director of MIT's CSAIL (Computer Science and Artificial Intelligence Laboratory.

Madhu Sudan's research interests include computational complexity theory, algorithms, and coding theory.  He is best known for his works on probabilistic checking  of proofs, and on the design of list-decoding algorithms for error-correcting codes.

In 2002, Madhu Sudan was awarded the Nevanlinna Prize, for outstanding contributions to the mathematics of computer science, at the International Congress of Mathematicians in Beijing. Madhu Sudan's other awards include the ACM Doctoral Dissertation Award (1992), the IEEE Information Theory Society Paper Award (2000) and the Godel Prize (2001), Distinguished Alumnus Award of the University of California at Berkeley (2003), and Distinguished Alumnus Award of the Indian Institute of Technology at Delhi (2004). Madhu Sudan is a Guggenheim Fellow (inducted 2005) and an ACM Fellow (inducted 2008).

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Colloquium/Architecture

Speaker/Affiliation:  Srini Devadas/MIT

Date/Time:  Monday, April 20, 2009  2:00 p.m.

Location:  ACES 2.402

Host:  Derek Chiou

Talk Title: "A Search for an Efficient Reconfigurable Computing Substrate"

Talk Abstract:

Computing substrates such as multicore processor chips or Field Programmable Gate Arrays (FPGAs) share the characteristic of having two-dimensional arrays of processing elements interconnected by a routing fabric.  At one end of the spectrum, FPGAs have a computing element that is a single-output programmable logic function and a statically-configurable network of wires.  At the other end, the computing element in a multicore is a complex 32-bit processor, and processors are interconnected using a packet-switched network. 

We are designing a reconfigurable substrate that shares characteristics of both FPGAs and multicores.  Our substrate is configured to run one application at a time, as with FPGAs. The computing element is a processor, and processors are connected using an interconnection network with virtual channel routers that use table-based routing. Bandwidth-sensitive oblivious routing methods that statically allocate virtual channels to application flows utilize the network efficiently. To accommodate bursty flows, the network contains adaptive bidirectional links that increase bandwidth in one direction at the expense of another. We are in the process of building a compiler that compiles applications onto this architecture so as to maximize average throughput of the applications. Our plan is to use the compiler to refine the architecture and then to build a reconfigurable processor chip.

Speaker Bio:

Srini Devadas is a Professor of Electrical Engineering and Computer Science at the Massachusetts Institute of Technology (MIT), and has been on the faculty of MIT since 1988.  He currently serves as the Associate Head of Computer Science.  Devadas has worked in the areas of Computer-Aided Design, testing, formal verification, compilers for embedded processors, computer architecture, computer security, and computational biology and has co-authored numerous papers and books in these areas.  Devadas was elected a Fellow of the IEEE in 1998.

---------------------------------------------------------------------------------------------------------------------------------------------
The Computer Architecture Seminar Series is sponsored jointly by the Departments of Computer Science and Electrical & Computer Engineering and is supported by a grant from AMD.
---------------------------------------------------------------------------------------------------------------------------------------------
 

Type of Talk:  UTCS Corporate Connection

Speaker/Affiliation:  Jun Sawada/IBM Austin Research Laboratory

Date/Time:  Monday, April 27, 2009  1:00 p.m.

Location:  ACES 2.302

Host:  UTCS Friends of Computer Science (FoCS)

Talk Title:  "Challenges for Industrial Research Laboratory in Post-Scaling Age"

Talk Abstract:

This talk will be targeted to the graduate students who are  interested in working for industrial research laboratories.  The speaker will introduce IBM Research, its recent achievements, and most importantly its challenges ahead.  IBM Research conducts annual studies on the global technology outlook and challenges facing the industry in the foreseeable future.  The significant slowdown of the semiconductor technology scaling forces us to seek alternative approaches to improve system performances.  The advent of internet-scale datacenters also require new system architectures and software running on top of it.  We cover these technology trends and discuss important research topics.

After the talk, the speaker will provide one-to-one discussion with the students who are interested in career opportunities at IBM Research. Those who are interested should send email to sawada@us.ibm.com and make an appointment.

Speaker Bio:

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Faculty Candidate

Speaker/Affiliation:  Luis Sentis/Stanford University

Date/Time:  Tuesday, April 21, 2009  11:00 a.m.

Location:  ACES 2.302

Host:  Risto Miikkulainen

Talk Title:  "Modeling and Control of Complex Skills in Humanoid Robots and Multi-Robot Systems"

Talk Abstract:

With the increasing expectation on autonomy for service and assistive robotics, the modeling and design of advanced robotic skills involving manipulation, locomotion, and behavior coordination in human environments are becoming essential to meet the emands.   General purpose helper robots and human assistive technologies depend on such critical advancements to be able to support and automate our daily chores.

My work focuses on the study and development of models and control methods of high-dimensional systems such as humanoid robots and large scale multi-vehicle systems operating in constrained environments. Highly dimensional systems such as humanoids are difficult to model and control because they involve the coordination of many low-level skills while fulfilling geometrical, contact and stability constraints imposed by the environment they operate in.  Likewise, large scale multi-robot systems such as heterogeneous vehicles operating in large geographical areas are difficult to coordinate because they involve the control of multiple operating criteria while maintaining network, geographical and formation constraints.

In this talk I will describe my approach to these problems addressing a model-based multi objective hierarchical control architecture that I have developed and matured over the years to create complex skills in the Asimo humanoid robot. As part of this framework, I will describe my work on modeling humanoid multi-contact behaviors to enable the synthesis of advanced manipulation and locomotion skills as well as the planning of interactive behaviors. If time allows, I will also describe some of the applications for multi-robot coordination in outdoor environments.

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Colloquium/Architecture

Speaker/Affiliation:  Mike Kistler/IBM Austin Research Laboratory

Date/Time:  Monday, April 27, 2009  3:30 p.m.

Location: ACES 2.302

Host:  Steve Keckler

Talk Title:  "Petascale Computing with Accelerators"

Talk Abstract:
A trend is developing in high performance computing in which commodity processors are coupled to various types of computational accelerators.  Such systems are commonly called hybrid systems. In this talk, I will describe our experience developing an implementation of the Linpack benchmark for a petascale hybrid system, the LANL Roadrunner cluster built by IBM for Los Alamos National Laboratory.  This system combines traditional x86-64 host processors with IBM PowerXCell™ 8i accelerator processors.  The implementation of Linpack we developed was the first to achieve a performance result in excess of 1.0 PFLOPS, and made Roadrunner the #1 system on the Top500 list in June 2008. I will describe the design and implementation of hybrid Linpack, including the special optimizations we developed for this hybrid architecture.  I will show actual results for single node and multi-node executions. From this work, we conclude that it is possible to achieve high performance for certain applications on hybrid architectures when careful attention is given to efficient use of memory bandwidth, scheduling of data movement between the host and accelerator memories, and proper distribution of work between the host and accelerator processors.

Speaker Bio:
Mike Kistler is a Senior Technical Staff Member in the IBM Austin Research Laboratory. He received his BA in Math and Computer Science from Susquehanna University in 1982, an MS in Computer Science from Syracuse University in 1990, and MBA from Stern School of Business (NYU) in 1991. He joined IBM in 1982 and has held technical and management positions in MVS, OS/2, and Lotus Notes development. He joined the IBM Austin Research Laboratory in May 2000 and is currently working on design and performance analysis for IBM's PowerPC and Cell/B.E. processors and systems. His research interests are parallel and cluster computing, fault tolerance, and full system simulation of high-performance computing systems.

Type of Talk:  UTCS Colloquium

Speaker/Affiliation:  Ilya Mironov/Microsoft Research Silicon Valley

Date/Time:  Thursday, April 30, 2009  3:30 p.m.

Location:  ACES 2.302

Host:  Vitaly Shmatikov

Talk Title:  "Building Privacy into the Netflix Prize Contenders"

Talk Abstract:

A recommender system based on collaborative filtering is a double-edged sword. By aggregating and processing preferences of multiple users it may provide relevant recommendations, serving an important business purpose. On the flip side, a recommender system is a potential source of leakage of private information shared by its users.

In the talk we will define differential privacy for recommender systems and briefly survey leading approaches in the Netflix Prize competition. We will demonstrate how these approaches can be adapted to provide differential privacy, without significantly degrading their accuracy.

Joint work with Frank McSherry, Microsoft Research.

Speaker Bio:

Ilya Mironov is a researcher at Microsoft Research, Silicon Valley Campus. He works in the areas of cryptography, cryptanalysis, and privacy of statistical databases. Ilya received his PhD from Stanford in 2003, advised by Dan Boneh. In his thesis he cryptanalyzed the popular RC4 stream cipher modeling it as a card shuffling scheme.
 

There is a signup schedule for this event (UT EID required).

Type of Talk:  UTCS Colloquium/Architecture

Speaker/Affiliation:  Lieven Eeckhout/Ghent University, Belgium

Date/Time:  Thursday, April 30, 2009   3:30 p.m.

Location:  ACES 2.402

Host:  Lizy John

Talk Title:  "Per-Thread Cycle Accounting in SMT Processors"

Talk Abstract:

Simultaneous Multi-threading (SMT) processors run multiple hardware threads simultaneously on a single processor core. While this improves hardware utilization substantially, co-executing threads affect each other's performance in often unpredictable ways. System software however is unaware of these performance interactions at the micro-architecture level, which may lead to unfair scheduling at the system level.

Starting from a mechanistic performance model, we derive a cycle accounting architecture for Simultaneous Multithreading (SMT) processors that estimates the execution times for each of the threads had they been executed alone, while they are running simultaneously on the SMT processor. This is done by accounting each cycle to either a base, miss event or waiting cycle component. Single-threaded alone execution time is then estimated as the sum of the base and miss event components; the waiting cycle component represents the lost cycle count due to SMT execution. The cycle accounting architecture incurs reasonable hardware cost (around 1KB of storage) and estimates single-threaded
performance accurately with average prediction errors around 7.2% for two-program workloads and 11.7% for four-program workloads.

The cycle accounting architecture has several important applications to system software and its interaction with SMT hardware. For one, the estimated single-thread alone execution time provides an accurate picture to system software of the actually consumed processor cycles per thread. The alone execution time instead of the total execution time (timeslice) may make system software scheduling policies more effective. Second, a new class of thread-progress aware SMT fetch policies based on per-thread progress indicators enable system software level priorities to be enforced at the hardware level. Third, per-thread cycle accounting enables substantially more effective symbiotic job scheduling.

Speaker Bio:
Lieven Eeckhout is an assistant professor at Ghent University, Belgium, and is a postdoctoral fellow with the Fund for Scientific Research -- Flanders (FWO). He received his PhD degree in computer science and engineering from Ghent University in 2002. His main research interest include computer architecture, virtual machines, performance modeling and analysis, simulation methodology, and workload characterization. He has published papers in top conferences such as ISCA, ASPLOS, HPCA, OOPSLA, PACT, CGO, DAC and DATE; he has served on multiple program committees including ISCA, PLDI, HPCA and IEEE Micro Top Picks; and he is the program chair for ISPASS 2009. His work on hardware
performance counter architectures was selected by IEEE Micro Top Picks from 2006 Computer Architecture Conferences as one of the "most significant research publications in computer architecture based on novelty and industry relevance". He graduated 5 PhD students, and currently supervises one postdoctoral researcher, 4 PhD students and 3 MSc students.