- talks - lists.cs.princeton.edu

PICASso TODAY: Analyzing Protein Interaction Networks
by Chrissy Colvin 29 Nov '06

29 Nov '06

** PICASso: ** Program in Integrative Information, Computer and Application Sciences ** www.cs.princeton.edu/picasso ** ** TALK TODAY--Wednesday, November 29, 2006 Computation and Data Analysis in Biology and Information Sciences www.cs.princeton.edu/picasso/bio_info_lunch.php TITLE: Analyzing Protein Interaction Networks SPEAKER: Mona Singh, Computer Science, Lewis-Sigler Institute, Princeton University TIME: TODAY Wednesday, November 29, 2006 Seminar begins at 12:30 p.m. (lunch provided 12:20) LOCATION: Room 402 Computer Science, Princeton University ABSTRACT: High-throughput technologies have determined physical interactions for a significant fraction of proteins in several organisms. Taken together, these linkages form large protein interaction networks. Computational analysis of these networks can help uncover the functions of individual proteins as well their roles within larger pathways. In this talk, I will introduce novel methods for analyzing protein interaction networks and uncovering recurring means with which diverse biological processes are carried out. Our approach lends insight into the organizational principles of protein networks and can be used to determine functionally cohesive subnetworks and help predict protein function. SIGN UP FOR THE PICASSO MAILING LIST: ===================================== If you would like to be kept informed of computationally-oriented events in (and around) Princeton, please SUBSCRIBE to the PICASso mailing list by visiting https://lists.cs.princeton.edu/mailman/listinfo/picasso. This page also contains information on how to UNSUBSCRIBE. PLEASE FORWARD THIS MESSAGE TO OTHER COMPUTATIONALLY-ORIENTED RESEARCHERS WHO MAY BE INTERESTED IN THIS EVENT, OR FUTURE PROGRAMS. THANKS!

1 0

D Nehab preFPO
by Melissa M Lawson 27 Nov '06

27 Nov '06

Diego Nehab will present his preFPO on Thursday November 30 at 4PM in Room 401 (note room!). The members of his committee are: Szymon Rusinkiewicz, advisor, James Davis (UC Santa Cruz), Tom Funkhouser, Adam Finkelstein, and Kai Li. Everyone is invited to attend his talk. His abstract appears below. --------------------------- In this talk, we present a variety of results that push the state of the art in 3D scanning techniques towards new simplicity, generality, and precision levels. We start with the introduction of the temporal stereo triangulation scanner, which is composed of two cameras and a projector. The system is substantially simpler than previous designs, because the projector does not need to be calibrated or even synchronized with the stereo camera pair. In order to improve the quality of our scanner, we introduce a novel algorithm that performs symmetric sub pixel refinement of integer matches. By simultaneously updating the matching coordinates on each camera, the method avoids biases that would otherwise appear as coherent noise in the final reconstructions. We then investigate the combination of data coming from a variety of different scanner types. We notice that the noise properties of acquired normal fields are generally unrelated to those commonly present in triangulated data. By carefully combining the best portion of each measurement type, we are able to achieve substantially higher quality results. Both normal acquisition and stereo triangulation are usually designed under the assumption of diffuse, opaque materials. We therefore conclude the talk with an investigation of alternative techniques for acquiring shape and orientation from translucent and/or highly specular materials.

1 0

Y Wang preFPO
by Melissa M Lawson 27 Nov '06

27 Nov '06

Yong Wang will present his preFPO on Thursday November 30 at 10:30AM in Room 402. The members of his committee are: Margaret Martonosi, advisor; Li-Shiuan Peh, Jennifer Rexford, Larry Peterson, and Kai Li. Everyone is invited to attend his talk.. His abstract follows below. ------------------------------ Wireless sensor networks and disruption-tolerant networks are new types of challenged networks that have very unpredictable and unusual situations, quite distinct from what a protocol architecture is usually designed for. The unique communication characteristics of these networks have created new, interesting challenges for the research community, such as opportunistic mobility, sparse connectivity, frequent network disruption/disconnection, and severely constrained energy budgets. Such challenges demand novel solutions to maintain efficient routing and take advantage of existing network services. This requires the network to be aware of such challenges and respond appropriately. In this thesis, I explore new optimizations via situation-awareness to improve routing performance in the presence of such challenges and disruptions. In particular, I study five categories of challenges that cover a wide range of disruptions experienced in challenged networks: sparse end-to-end connectivity, highly varying mobility, opportunistic connectivity, intermittent connectivity, and network congestion. The first three categories are confined to mobile networks, while the last two categories apply to static networks. The first category consists of networks with sparse end-to-end connectivity. Therefore, route breakage may bring a communication pair into temporary disconnection or lead to nodes using stale routes. I propose a timekeeping technique to drive route prefetching and decay operations based on route lifetime information. Through extensive simulations, I demonstrate up to 31% improvement in packet delivery rate and 53% improvement in average end-to-end packet latency. The second category consists of networks with a varying mobility pattern that is typical in many emerging networks, an example of which is ZebraNet. With such a mobility pattern, nodes move in phases that have very different characteristics I propose a model-based approach to capture mobility phase changes in order to maintain efficient routing. When evaluated using a mobility trace from the real-world ZebraNet deployment, the situation-aware approach leads to an improvement of up to 120% in packet delivery rate. _______________________________________________ talks mailing list talks(a)lists.cs.princeton.edu https://lists.cs.princeton.edu/mailman/listinfo/talks

1 0

Y Wang preFPO
by Melissa M Lawson 27 Nov '06

27 Nov '06

Young Wang will present his preFPO on Thursday November 30 at 10:30AM in Room 402. The members of his committee are: Margaret Martonosi, advisor; Li-Shiuan Peh, Jennifer Rexford, Larry Peterson, and Kai Li. Everyone is invited to attend his talk.. His abstract follows below. ------------------------------ Wireless sensor networks and disruption-tolerant networks are new types of challenged networks that have very unpredictable and unusual situations, quite distinct from what a protocol architecture is usually designed for. The unique communication characteristics of these networks have created new, interesting challenges for the research community, such as opportunistic mobility, sparse connectivity, frequent network disruption/disconnection, and severely constrained energy budgets. Such challenges demand novel solutions to maintain efficient routing and take advantage of existing network services. This requires the network to be aware of such challenges and respond appropriately. In this thesis, I explore new optimizations via situation-awareness to improve routing performance in the presence of such challenges and disruptions. In particular, I study five categories of challenges that cover a wide range of disruptions experienced in challenged networks: sparse end-to-end connectivity, highly varying mobility, opportunistic connectivity, intermittent connectivity, and network congestion. The first three categories are confined to mobile networks, while the last two categories apply to static networks. The first category consists of networks with sparse end-to-end connectivity. Therefore, route breakage may bring a communication pair into temporary disconnection or lead to nodes using stale routes. I propose a timekeeping technique to drive route prefetching and decay operations based on route lifetime information. Through extensive simulations, I demonstrate up to 31% improvement in packet delivery rate and 53% improvement in average end-to-end packet latency. The second category consists of networks with a varying mobility pattern that is typical in many emerging networks, an example of which is ZebraNet. With such a mobility pattern, nodes move in phases that have very different characteristics I propose a model-based approach to capture mobility phase changes in order to maintain efficient routing. When evaluated using a mobility trace from the real-world ZebraNet deployment, the situation-aware approach leads to an improvement of up to 120% in packet delivery rate.

1 0

PLOrk performance tonight
by Ge Wang 21 Nov '06

21 Nov '06

Dear all, This event won't be a talk (we hope), but it seemed more relevant to announce here than to [building] or [downtime]. http://ffmup.org/nov_21_06.html --- who/what: The Princeton Laptop Orchestra (PLOrk) presents an evening of new works for computer-mediated chamber music (duo/trio/quartet), composed and performed by members of the Fall 2006 ensemble and seminar: Alan Tormey | Ananya Misra | Andrea Mazzariello | Anne Hege | Ge Wang | John Fontein | Laurie Hollander | Matt Hoffman | Margaret R. Daniel | Michael Hammond | Perry Cook | Rebecca Fiebrink | Samson Young | Scott Elmegreen | Scott Smallwood | Spencer Salazar --- when: tonight 9:00pm (Tuesday) where: Terrace Club (free admission) why: there is usually free beer at ffmup events; new music; free beer! Hope to see you there! Best, Ge!

1 0

Michael J Freedman, Mon Nov 20, 1:30
by Vivek Pai 16 Nov '06

16 Nov '06

Michael J Freedman, the NYU graduate student, will be speaking on Monday at 1:30 in room 402. His talk title and abstract are included below. If you'd like to talk with him, please sign up in /n/fs/department/schedules/visitors/visitfall06/freedman Just to avoid some confusion that occurred the last time he visited, please note that this is not Michael H. Freedman, the Fields Medal winner. =================== RE: Fighting Spam with Reliability and Privacy Mike Freedman NYU / Stanford http://www.scs.stanford.edu/mfreed/ mfreed(a)scs.stanford.edu ----- The explosive growth in unwanted email has prompted the development of techniques for the rejection of email, intended to shield recipients from the onerous task of identifying the legitimate email in their inboxes amid a sea of spam. Unfortunately, widely-used content-based filtering systems have converted the spam problem into a false positive one: Email has become unreliable. We present Reliable Email (RE: [1]), a new whitelisting system that incurs zero false positives among socially-connected users yet populates whitelists automatically by exploiting transitive trust relationships. Further, while RE: permits a recipient to discover whether other users have whitelisted the sender, it preserves the privacy of email contacts with a cryptographic Private Matching [2] protocol. Using real email traces from two sites, we demonstrate that RE: renders a significant fraction of received email reliable. Our evaluation also shows that RE: can prevent up to 88% of the false positives incurred by common email rejection systems. Finally, we propose a more formal definition of security for verifying social connectedness in a privacy-preserving fashion [3]. In the process, we identify a mismatch between RE:'s goals and [2]'s properties, enabling us to construct new cryptographic protocols that are either more efficient (based only on symmetric operations) or achieve stronger guarantees (namely, non-transferability) through bilinear pairings. Using such, RE: can provide privacy-preserving social whitelisting with millisecond overhead. 1. Re: Reliable Email Garriss, Kaminsky, Freedman, Karp, Mazieres, and Yu, NSDI '06 2. Efficient Private Matching and Set Intersection Freedman, Nissim, and Pinkas, EUROCRYPT '04 3. Efficient Private Techniques for Verifying Social Proximity Freedman and Nicolosi, Submitted Nov '06 ----- Michael J. Freedman is a doctoral student at NYU, currently visiting Stanford University. He received his M.Eng. and S.B. degrees from MIT. His research interests broadly focus on security, distributed systems, and cryptography. Among other things, he developed the Coral content distribution network and OASIS anycast service, which serve more than a million users daily (see www.coralcdn.org). In 2006, he co-founded Illuminics Systems, which has been acquired by Quova, Inc.

1 0

Gödel Centenary Program this Friday
by Moses Charikar 15 Nov '06

15 Nov '06

On Friday, November 17, from 2:00 p.m. to 6:30 p.m. in Wolfensohn Hall, the Institute will host A Program to Mark the Centenary Year of the Birth of Kurt Gödel. The event is free and open to the public; reservations are not required. Seating is on a first come, first served basis. The program will consist of talks aimed at a general audience on the life and work of Kurt Gödel (1906-1978) and his impact on mathematics, philosophy, and computer science. Participants include: John Burgess, Princeton University; John W. Dawson, Jr., The Pennsylvania State University; Solomon Feferman, Stanford University; Juliette Kennedy, University of Helsinki; Karl Sigmund, University of Vienna; and Avi Wigderson, Institute for Advanced Study. For additional information about the program, please visit http://www.ias.edu/SpFeatures/kurt_godel/godel-4.php. There is also background on Gödel's life and work accessible from the IAS homepage, www.ias.edu. An exhibition of posters on Gödel, created by the University of Vienna, will remain on view on the main floor of Fuld Hall through Sunday, December 3. --Moses

1 0

Talk Today, 4pm - CS 105 - Serafim Batzoglou
by Ginny Hogan 07 Nov '06

07 Nov '06

Serafim Batzoglou will present his talk "Models and Algorithms for Genomic Sequences, Proteins, and Networks of Protein Interactions" Wednesday Nov 8th at 4:00pm Computer Science Room 105.

1 0

manfred warmuth talk at 4pm today
by Robert Schapire 25 Oct '06

25 Oct '06

Manfred Warmuth from UC Santa Cruz will be giving today's colloquium at 4pm in CS room 105. See abstract below, or visit http://www.cs.princeton.edu/research/colloquia.php Rob Schapire ------------------------------------------------------------------------ *Leaving the Span* *Manfred K. Warmuth* University of California, Santa Cruz When linear models are too simple then the following "kernel trick" is commonly used: Expand the instances into a high-dimensional feature space and use any algorithm whose linear weight vector in feature space is a linear combination of the expanded instances. Linear models in feature space are typically non-linear in the original space and seemingly more powerful. Also dot products can still be computed efficiently via the use of a kernel function. However we discuss a simple sparse linear problem that is hard to learn with any algorithm that uses a linear combination of the embedded training instances as its weight vector, no matter what embedding is used. We show that these algorithms are inherently limited by the fact that after seeing /k/ instances only a weight space of dimension /k/ can be spanned. Surprisingly the same problem can be efficiently learned using the exponentiated gradient (EG) algorithm: Now the component-wise logarithms of the weights are essentially a linear combination of the training instances. This algorithm enforces "additional constraints" on the weights (all must be non-negative and sum to one) and in some cases these constraints alone force the rank of the weight space to grow as fast as /2^k /. (Joint work with S.V.N. Vishwanathan!) ------------------------------------------------------------------------ -- Princeton University Department of Computer Science 35 Olden Street Princeton, NJ 08540 USA tel: +1 609 258 7726 fax: +1 609 258 1771 Computer Science Building, Room 407 http://www.cs.princeton.edu/~schapire schapire(a)cs.princeton.edu

1 0

talk by Yoav Freund -- Friday, 7/14, 2:30, 402 CS
by Robert Schapire 11 Jul '06

11 Jul '06

Yoav Freund will be giving a talk this Friday, July 14 at 2:30pm in room 402 of the CS building. Title and abstract follow below. ------------------------------------------------------------------------ Title: Using adaptive computer vision to analyze biomedical images Speaker: Yoav Freund Computer Science and Engineering, University of California, San Diego http://www.cse.ucsd.edu/~yfreund Date: Friday, July 14, 2:30pm Place: Room 402, CS building Host: Rob Schapire Abstract: The rapid advancement of biomedical imaging techniques such as fluorescence microscopy, non-linear optics, MRI and ultra-sound make it now possible to observe the time evolution of mechanical and biochemical processes in cells and in the human body at a resolution unimaginable a decade ago. This ability to measure opens up an incredible range of new possibilities for research in biology and in medicine. However, a major challenge along the way is the formidable amount of manual work required to extract meaningful quantitative results from the enormous amounts of collected image data. I suggest a combination of computer vision and machine learning techniques to tackle such problems. In this talk I describe some of my current work applying these ideas in several projects: 1. The study of the dynamics of lamellipodia in spreading cells. 2. The study of regulatory networks in the development of drosophila embryos. 3. Detection of protein micro-crystals for X-ray crystalography. 4. Detection and grading of prostate cancer using microscope images of biopsies. I will conclude with some of the lessons I have learned from these projects and some words on my current approach to automating image analysis tasks. ------------------------------------------------------------------------ -- Princeton University Department of Computer Science 35 Olden Street Princeton, NJ 08540 USA tel: +1 609 258 7726 fax: +1 609 258 1771 Computer Science Building, Room 407 http://www.cs.princeton.edu/~schapire schapire(a)cs.princeton.edu

1 0