- talks - lists.cs.princeton.edu

EE Seminar- Wed., April 14 12pm- Yuan Xie
by Nicole E. Wagenblast 12 Apr '10

12 Apr '10

Department of Electrical Engineering Computer Engineering Seminar Speaker: Prof. Yuan Xie The Pennsylvania State University Date: April 14, 2010 Time: 12:00 p.m. (Pizza will be served.) Room: E-Quad, Room B327 Title: Emerging 3D Stacking and Non-volatile memory technologies and their impact on Computer Design The growing complexity of advanced deep-submicron technology used to build computing systems has resulted in many design challenges for future computing systems, such as power limits, limited communication bandwidth and latency, and reduced system reliability. At the same time, various disruptive emerging technologies promise dramatically improved performance at the device level. How likely are these improvements and what impact will they have at the circuit and architecture level? In this talk, I will present an overview of two emerging technologies: 3D integration and new non-volatile memory technologies. I will describe the fundamentals and current status of each technology, introduce the advantages of the technologies, and discuss the design challenges inherent in adoption of the technologies, and give an overview of the research activities in my group. A preview of these two research areas can be found at http://www.cse.psu.edu/~yuanxie/3d.html and http://www.cse.psu.edu/~yuanxie/nvm.html Biography: Prof. Yuan Xie received the B.S. degree from Tsinghua University and the M.S. and Ph.D. degrees in electrical engineering from Princeton University. He is currently Associate Professor in CSE department at the Pennsylvania State University. Before joining Penn State in Fall 2003, he was with IBM Microelectronic Division's Worldwide Design Center. Prof. Xie is a recipient of the NSF CAREER award, the SRC Inventor Recognition Award, IBM Faculty Award, several Best Paper Award and Best Paper Award Nominations at IEEE/ACM conferences. He has published more than 100 papers in journals and refereed conference proceedings, in the area of EDA, computer architecture, VLSI circuit designs, and embedded systems. His current research projects focus on 3D IC design and emerging memory technologies. He is currently Associate Editor for ACM Journal of Emerging Technologies in Computing Systems (JETC), IEEE Transactions on Very Large Scale Integration Systems (TVLSI), IEEE Transactions on Computer Aided Design of Integrated Circuits (TCAD), IEEE Design and Test of Computers, IET Computers and Digital Techniques (IET CDT). He is ACM Distinguished Speaker. He is also a co-editor of the first 3D IC book on EDA, Design and Architecture (Springer 2009). Host: Prof. Niraj Jha

1 0

Colloquium Speaker, Wed April 14- Diana Young
by Nicole E. Wagenblast 12 Apr '10

12 Apr '10

Measuring Human Motion: Applications in Music Interfaces and Assistive Devices Diana Young, MIT Media Lab Wednesday, April 14, 2010 4:30pm Small Auditorium, CS105 Advances in sensing and computing capabilities of embedded systems allow accurate and precise measurement of human motion, enabling studies of diverse physical tasks and the development of sophisticated human-computer interfaces driven by gestural input. Leveraging prior work in the development of interactive music controllers, a wireless measurement system comprised of inertial, force, and position sensors, is implemented on a violin and bow in order to capture the primary bowing parameters (bow force, bow speed, and bow-bridge distance) produced during realistic performances. A study of standard bowing techniques performed by expert violinists shows that with appropriate data reduction and pattern recognition techniques, gesture data from the above sensors may be used to recognize the same technique when performed by different players, as well as to begin to discriminate different players' performances of the same techniques. Accurate interpretation of human motion is not only useful for understanding expressive musical performance, but it is crucial for other human-computer interfaces, such as assistive mobility devices, in which ensuring safety is a chief concern. New lower limb prostheses and orthoses designed to correct gait pathologies (caused by disease or injury) include active elements that must be controlled in realtime as a function of the user's motion. Using sensing systems similar to that described above, relevant gait parameters (such as orientation, velocity, and position) may be estimated for use as control inputs for these active devices, as well as for general use in the study of gait biomechanics. Results and current progress will be presented, and implications for future applications in new music instrument development, music performance pedagogies, and interventions for rehabilitation and augmentation will be discussed. Diana Young is a Postdoctoral Associate in the Biomechatronics research group at the MIT Media Lab, where she contributes to the development of measurement systems that capture human locomotion for use in active lower limb prostheses. Her research interests include measurement and understanding of human motion, both functional and expressive, and the development of human-computer interfaces for rehabilitation and the enhancement of skill acquisition. For her doctoral studies on the physical dynamics of violin bowing, she designed a playable calibrated measurement system to capture and record physical bowing parameters generated by players. She has also developed several related gesture sensing systems for use in live interactive music performance by professional musicians. Her work has been featured in numerous media, including The Strad magazine, Strings Magazine, The Boston Globe, New Scientist Podcast series, and Scientific American Frontiers. Young holds a Ph.D. and M.S. in Media Arts and Sciences from MIT, a B.S. in Electrical Engineering and B.A. in Music from The Johns Hopkins University, and a Performer's Certificate in Violin from the Peabody Conservatory of Music.

1 1

C Bienia preFPO
by Melissa Lawson 12 Apr '10

12 Apr '10

Christian Bienia will present his preFPO on Wednesday April 14 at 2PM in Room 402. The members of his committee are: Kai Li, advisor; J.P. Singh and David August, readers; Jen Rexford and Margaret Martonosi, nonreaders. Everyone is invited to attend his talk. His abstract follows below. ----------------------------------------- TITLE: Benchmarking Modern Multiprocessors ABSTRACT: Benchmarking has become one of the most important methods for quantitative performance evaluation of processor and computer system designs. Benchmarking of modern multiprocessors such as chip-multiprocessors is challenging because of their different application domain, scalability and parallelism requirements. In my thesis, I have developed a methodology to design effective benchmark suites and demonstrated its effectiveness by developing and deploying a benchmark suite for evaluating multiprocessors. More specifically, my thesis includes several contributions. First, the thesis shows that parallel program behavior has changed significantly over time and that new benchmark designs need to consider such changes. Second, the thesis quantitatively describes the requirements and characteristics of current multithreaded programs and the underlying technology trends that have shaped them. Third, the thesis presents a systematic approach to scale and select benchmark inputs with the goal of optimizing benchmarking accuracy subject to constrained execution or simulation time. Finally, the thesis describes how I developed a parallel benchmark suite called PARSEC for evaluating modern shared-memory multiprocessors. Since its initial release, PARSEC has been adopted by most architecture groups in both research and industry.

1 0

2pm talk today on content-centric opportunistic communication for mobile phones
by Jennifer Rexford 09 Apr '10

09 Apr '10

TALK ANNOUNCEMENT Speaker: Erik Nordstrom, Princeton University Title: Haggle: Content-Centric Opportunistic Communication for Mobile Phones Date/time: 2:00-3:00pm on Friday April 9 Location: room 402 in the Computer Science building Abstract: Today, mobile phones are increasingly used for content-centric communication, whereby users access and share content with each other (e.g., pictures on Facebook). However, content sharing is limited to infrastructure networks and legacy Internet technologies are ill suited for content-centric communication. Moreover, the Internet protocols and APIs do not work well in highly mobile and disruptive environments, where intermittent connectivity is the norm rather than the exception. These deficiencies make it is cumbersome, or even impossible, to share content directly between mobile phones, leading to high stress on existing infrastructure and high costs for the consumer, while putting limitations on the possibilities for developing new and innovative applications. In this talk, I will present Haggle, a novel network architecture built from the ground up to support opportunistic content-centric communication directly between mobile phones. Haggle addresses content rather than hosts, which makes it irrelevant from where a user receives a content item. The content-centric communication is driven by a novel distributed search-based resolution scheme, which draws inspiration from online search engines. The search scheme resolves the mappings between the content and its interest group, i.e., the nodes that have an interest in the content. Search has the benefit of ranking, and therefore provides a natural limitation and ordering mechanism for content dissemination. This means that content, in general, has a higher likelihood of being disseminated if there is a high interest in it in the network. After the talk, I will demonstrate Haggle using a visualization tool that shows how photos taken with a phone's camera are shared with other mobile phones or devices. See http://www.haggleproject.org/ for more details.

1 0

Colloquiym Speaker, Tues April 13th- Wojciech Matusik
by Nicole E. Wagenblast 08 Apr '10

08 Apr '10

Bridging the Gap to the Real Wojciech Matusik, Disney Research Zurich Tuesday, April 13th- 4:30pm Small Auditorium CS105 Devices and systems for measuring real-world shape, motion, and appearance have contributed to progress in computer graphics over the last 15 years. For example, digital photography and image-based rendering have had a profound impact on rendering. Motion capture has transformed computer animation. Geometry acquisition using 3D scanners has driven developments in geometry processing. In data-driven computer graphics, we have replaced hand-modeled and procedurally-defined components with real-world measurements. Furthermore, we have developed tools to modify and edit these data-driven representations. With these new methods in hand, computer graphics has succeeded in recreating reality on a computer screen. The next big challenge for computer graphics and interactive techniques is to expand beyond a computer display. In the next years, the field will be working on algorithms and tools that will allow us to design and automatically fabricate physical output. This trend will dramatically increase the scope of computer graphics and it will exert significant impact on our daily lives. In my talk, I will show two instances of a process that bridges the gap between the digital and real worlds. First, I will describe a mathematical modeling of light reflection from surfaces, based on measurements of physical materials, as well as the reverse process: manufacturing physical materials with desired reflectance properties. Second, I will present an analogous process for measurement, modeling, and fabrication of objects with desired deformation behavior. Not only do these research projects showcase how the scope of computer graphics methods can be greatly extended, but they also prompt fundamental research into the representation and processing of data types that go beyond images and video. Wojciech Matusik is a senior research scientist at Disney Research Zurich. He received a B.S. in EECS from the University of California at Berkeley in 1997, a M.S. in EECS from MIT in 2001, and a Ph.D. in 2003. In 2004, he was named one of the world's top 100 young innovators by MIT's Technology Review Magazine. He received an ACM SIGGRAPH Significant New Researcher Award in 2009. His primary research is in the field of computer graphics, with broad applications in other disciplines such as digital communications, materials science, optics, and biomechanics.

1 0

Colloquium Speaker, Mon April 12- Jake Abernethy
by Nicole E. Wagenblast 06 Apr '10

06 Apr '10

Learning, Adversaries, and Limited Feedback Jake Abernethy, UC Berkeley Monday, April 12, 2010 4:30pm Small Auditorium, CS 105 A basic assumption often made in Machine Learning is that the data we observe are independent and identically distributed. But is this really necessary? Is it even realistic in typical scenarios? A number of recent results provide guarantees for arbitrary (or even adversarially-generated) data sequences. It turns out, for a wide class of problems, the learner's performance is no worse when the data is assumed to be arbitrary as opposed to i.i.d. This result involves a nice application of the Minimax theorem, which I'll briefly describe. I'll also delve into a more challenging problem: the "bandit" setting, in which the learner receives only very little feedback on each example. It was unknown for some time whether there exists an efficient algorithm that achieves the same guarantee for non-i.i.d. data for a general class of learning/decision problems. I'll present the first known such bandit algorithm, and I'll sketch the central ideas behind the technique, borrowing several tricks from the Interior Point optimization literature.

1 0

2pm Fri Apr 9 talk on Content-Centric Opportunistic Communication for Mobile Phones
by Jennifer Rexford 05 Apr '10

05 Apr '10

TALK ANNOUNCEMENT Speaker: Erik Nordstrom, Princeton University Title: Haggle: Content-Centric Opportunistic Communication for Mobile Phones Date/time: 2:00-3:00pm on Friday April 9 Location: room 402 in the Computer Science building Abstract: Today, mobile phones are increasingly used for content-centric communication, whereby users access and share content with each other (e.g., pictures on Facebook). However, content sharing is limited to infrastructure networks and legacy Internet technologies are ill suited for content-centric communication. Moreover, the Internet protocols and APIs do not work well in highly mobile and disruptive environments, where intermittent connectivity is the norm rather than the exception. These deficiencies make it is cumbersome, or even impossible, to share content directly between mobile phones, leading to high stress on existing infrastructure and high costs for the consumer, while putting limitations on the possibilities for developing new and innovative applications. In this talk, I will present Haggle, a novel network architecture built from the ground up to support opportunistic content-centric communication directly between mobile phones. Haggle addresses content rather than hosts, which makes it irrelevant from where a user receives a content item. The content-centric communication is driven by a novel distributed search-based resolution scheme, which draws inspiration from online search engines. The search scheme resolves the mappings between the content and its interest group, i.e., the nodes that have an interest in the content. Search has the benefit of ranking, and therefore provides a natural limitation and ordering mechanism for content dissemination. This means that content, in general, has a higher likelihood of being disseminated if there is a high interest in it in the network. After the talk, I will demonstrate Haggle using a visualization tool that shows how photos taken with a phone's camera are shared with other mobile phones or devices. See http://www.haggleproject.org/ for more details.

1 0

Colloquium Speaker, Tues April 6- Rebecca Fiebrink
by Nicole E. Wagenblast 02 Apr '10

02 Apr '10

Real-time Human-Computer Interaction with Supervised Learning Algorithms for Music Composition and Performance Rebecca Fiebrink, Princeton University, Computer Science Department Tuesday, April 6, 2010- 4:30pm Small Auditorium CS105 Supervised learning offers a useful set of algorithmic tools for many problems in computer music composition and performance. Through the use of training examples, these algorithms offer human musicians a means to implicitly specify the relationship between low-level, human-generated control signals (such as gesturally-manipulated sensor outputs or audio captured by a microphone) and the desired computer response (such as a change in synthesis or structural parameters of dynamically-generated audio). In my work, I explore how to most effectively enable users to interact with supervised learning algorithms to compose and perform new music. I have built a general-purpose software system for applying standard supervised learning algorithms in real-time problem domains. This system, called the Wekinator, supports human interaction throughout the entire supervised learning process, including the generation of training examples and the application of trained models to real-time inputs. Already, the Wekinator has enabled the creation of several new compositions and instruments. Furthermore, this system has enabled me to study several aspects of human-computer interaction with supervised learning in computer music. I have used the Wekinator as a foundation for a participatory design process with practicing composers, ongoing work with non-expert users in a classroom context, and the design of a gesture recognition system for a sensor-augmented cello bow. This research has led to a clearer characterization of the requirements and goals of instrument builders and composers, a better understanding of how to design user interfaces for supervised learning in both real-time and creative application domains, and a greater insight into the roles that interaction (encompassing both human-computer control and computer-human feedback) can play in the development of systems containing supervised learning components. This work highlights how music and other creative endeavors differ from more traditional applications of supervised learning, and it contributes to a broader HCI perspective on machine learning practice.

1 1

C Richards FPO
by Melissa Lawson 01 Apr '10

01 Apr '10

Chris Richards will present his FPO on Thursday April 8 at 11AM in Room 402. The members of his committee are: Andrew Appel, advisor; David Walker and Paul-Andre Mellies (CNRS), readers; David August and Sanjeev Arora, nonreaders. A copy of his thesis, "The Approximation Modality in Models of Higher-Order Types" is on display in Room 310. Everyone is invited to attend his talk.

1 0

Korhammer Lecture: The Art of Green Design: Doing Nothing Well - Prof. Jan M. Rabaey, UC Berkeley
by Jennifer Rexford 25 Mar '10

25 Mar '10

> > > KORHAMMER LECTURE > > PRINCETON UNIVERSITY, DEPARTMENT OF ELECTRICAL ENGINEERING > > “The Art of Green Design: Doing Nothing Well” > > Jan M. Rabaey, Donald O. Pederson Distinguished Professor, University of California, Berkeley > > Monday, March 29, 2010, 4:30pm, Computer Science 104 > > Abstract: > Reducing human appetite for energy and the corresponding carbon footprint is one of the dominant societal problems of today. Information technology (IT) very likely will play a major role in our quest to generate, distribute, store and consume energy more efficiently. Unfortunately the IT platform itself represents a major consumer of electrical power, and its energy footprint is very likely to continue to grow over the next decades. > > The most obvious approach to address this challenge is to continue to improve the energy efficiency of our computation, communication or storage devices. Unfortunately, the opportunities of doing so (using technology scaling or innovative architectures) are getting increasingly limited. Instead, a lower-hanging fruit is to eliminate the massive amount of waste that is inherent to today’s IT systems. Systems tend to be designed and optimized for peak performance. In reality, most computation nodes, networks and storage devices typically operate at a fraction of the maximum load, and do this with surprisingly low energy efficiency. If we could design systems that do “nothing” well (as phrased by David Culler), major energy savings would be enabled. Accomplishing “energy-proportional” computing requires a full-fledged top-down and bottom-up approach to the design of IT systems. This talk discusses what this means for a couple of systems ranging from large scale data centers over communication networks to small-scale body-area networks for medical applications. > > > > >

1 0