February 2014 - talks - lists.cs.princeton.edu

Colloquium Speaker Barbara Engelhardt Monday, March 3, 4:30pm
by Nicole E. Wagenblast 03 Mar '14

03 Mar '14

Statistical models to study how a genetic variant impacts an organism Barbara Engelhardt ( Duke University ) Monday, March 3, 4:30pm Computer Science 105 Consider sequencing the genome of a newborn, and selecting targeted therapeutics early in life to reduce her lifetime risk of addiction, obesity, type II diabetes, or pancreatic cancer. While genome-wide association studies (GWAS) have unquestionably been successful in identifying reproducible genomic risk factors for complex human diseases, the promise of developing therapeutics to reduce the heritable portion of disease risk is far from fulfillment. The essential technological developments to fulfill this promise, however, are mainly in statistics and computation rather than in genomic experimental methods. I describe three genomic studies from my recent work. First, I identified a genetic variant that behaves differently depending on whether or not an individual takes cholesterol-reducing drugs. Second, I found that genetic variants that are associated with different cell traits are co-localized with a large variety of different regulatory mechanisms more often than expected. Third, I developed a model to uncover genetic variants that affect many traits simultaneously, where the trait measurements have substantial technical noise. Throughout, I emphasize statistical and computational challenges, and innovations necessary to fulfill this promise of genomic studies.

1 1

Colloquium Speaker Anima Anandkumar Thurs Feb 27, 4:30pm
by Nicole E. Wagenblast 27 Feb '14

27 Feb '14

Guaranteed Learning of Latent Variable Models: Overlapping Community Models and Overcomplete Representations Anima Anandkumar , University of California Irvine Thursday, February 27, 4:30pm Computer Science 105 Incorporating latent or hidden variables is a crucial aspect of statistical modeling. I will present a statistical and a computational framework for guaranteed learning of a wide range of latent variable models. I will focus on two instances, viz., community detection and overcomplete representations. The goal of community detection is to discover hidden communities from graph data. I will present a tensor decomposition approach for learning probabilistic mixed membership models. The tensor approach is guaranteed to correctly recover the mixed membership communities with tight guarantees. We have deployed it on many real-world networks, e.g. Facebook, Yelp and DBLP. It is easily parallelizable, and is orders of magnitude faster than the state-of-art stochastic variational approach. I will then discuss recent results on learning overcomplete latent representations, where the latent dimensionality can far exceed the observed dimensionality. I will present two frameworks, viz., sparse coding and sparse topic modeling. Identifiability and efficient learning are established under some natural conditions such as incoherent dictionaries or persistent topics. Anima Anandkumar is a faculty at the EECS Dept. at U.C.Irvine since August 2010. Her research interests are in the area of large-scale machine learning and high-dimensional statistics. She received her B.Tech in Electrical Engineering from IIT Madras in 2004 and her PhD from Cornell University in 2009. She has been a visiting faculty at Microsoft Research New England in 2012 and a postdoctoral researcher at the Stochastic Systems Group at MIT between 2009-2010. She is the recipient of the Microsoft Faculty Fellowship, ARO Young Investigator Award, NSF CAREER Award, IBM Fran Allen PhD fellowship, thesis award from ACM SIGMETRICS society, paper awards from the ACM SIGMETRICS and IEEE Signal Processing societies, and 2014 Sloan Fellowship.

1 1

Colloquium Speaker Xi Wang, Wed Feb 26, 4:30pm
by Nicole E. Wagenblast 26 Feb '14

26 Feb '14

A new class of bugs: How compiler optimizations harm our systems through undefined behavior Xi Wang , Massachusetts Institute of Technology Wednesday, February 26, 4:30pm Computer Science 105 Software bugs introduce security vulnerabilities into our computer systems. To understand and mitigate an increasing number of bugs, practitioners categorize them into classes, such as buffer overflow or SQL injection, and handle each class separately. This talk introduces a new class of bugs called unstable code: code that is unexpectedly discarded by compiler optimizations due to undefined behavior in the program. I will discuss its prevalence and security impact in systems, and present a systematic approach for reasoning about unstable code, as well as a static checker called Stack that implements this approach to precisely identify unstable code in real systems. Applying Stack to widely used software has uncovered 160 new bugs that have been confirmed and fixed by developers. It has also been adopted by several companies to scan their codebases. Xi Wang is a PhD candidate in Computer Science at MIT, advised by M. Frans Kaashoek and Nickolai Zeldovich. His research interests are in building secure and reliable systems. He was awarded a Best Paper Award at SOSP 2013, a Best Student Paper Award at EuroSys 2008, and an MIT Jacobs Presidential Fellowship in 2008.

2 2

Colloquium Speaker Rishabh Singh Tues Feb 25, 4:30pm
by Nicole E. Wagenblast 25 Feb '14

25 Feb '14

Program Synthesis for the Masses Rishabh Singh ( Massachusetts Institute of Technology ) Tuesday, February 25, 4:30pm Computer Science 105 New computing platforms have greatly increased the demand for programmers, but learning to program remains a big challenge. Program synthesis has the potential to revolutionize programming by making it more accessible. My work has focused on two goals: making programming more intuitive through the use of new interfaces, and using automated feedback to help students learn programming. In this talk, I will present my work on three systems that work towards these goals. The FlashFill system helps end-users perform repetitive data transformations over strings, numbers, and tables using input-output examples. FlashFill was shipped as part of Excel 2013 and was quoted as one of the top features by many press reviews. The Storyboard Programming system helps students write data-structure manipulations using textbook-like visual examples and bridges the gap between high-level insights and low-level code. Finally, the Autograder system provides automated feedback to students on introductory programming assignments, and was successfully run on tens of thousands of programming exercises from edX. I will describe how ideas from advances in constraint-solving, machine learning, and formal verification enabled the new forms of interaction required by these systems. Rishabh Singh is a PhD candidate in the Computer Science and Artificial Intelligence Laboratory at MIT. His research interests are broadly in formal methods and programming languages. His PhD work focuses on developing program synthesis techniques for making programming accessible to end-users and students. He is a Microsoft Research PhD fellow and winner of MIT's William A. Martin Outstanding Master's thesis Award. He obtained his BTech in Computer Science and Engineering from IIT Kharagpur in 2008, where he was awarded the Institute Silver Medal and Bigyan Sinha Memorial Award. He was also awarded to be Prime Minister's National Guest at Republic Day Parade, New Delhi in 2005.

1 2

Princeton Public Lectures: Google's Eric Schmidt on campus 2/26/2014
by Nicole E. Wagenblast 21 Feb '14

21 Feb '14

Flyer attached.

1 0

4:30pm Wed Mar 12 talk by Cerf and Kahn on the evolution of the Internet
by Jennifer Rexford 20 Feb '14

20 Feb '14

Speakers: Vinton Cerf, Chief Internet Evangelist, Google Robert Kahn, CEO of Corporation for National Research Initiatives Title: Fathers of the Internet Reflecting on the Evolution of the Internet Day: 4:30pm Wed March 12 Room: Friend Center Auditorium 101 Host: Mung Chiang Vinton Cerf (Chief Internet Evangelist at Google) and Robert Kahn (CEO of Corporation for National Research Initiatives), whose pioneering work on the Internet has been widely recognized by numerous awards, including the US Presidential Medal of Freedom, the US National Medal of Technology, and the ACM Turing Award, will together give a public duet lecture on “Perspectives on the Internet and its Evolution." This is an open lecture to the general public. No fee or ticket required. Seating available on a first-come-first serve basis. Seating will begin at 4pm. Reception will follow. Additional information available at http://edgelab-events.princeton.edu/

1 0

Colloquium Speaker Aaron Turon Thurs Feb 20, 4:30pm
by Nicole E. Wagenblast 19 Feb '14

19 Feb '14

Practical Foundations for Scalable Concurrency Aaron Turon ( Max Planck Institute for Software Systems (MPI-SWS) Thursday, February 20, 4:30pm Computer Science 105 The traditional way to tame concurrency is to disallow it: "critical sections" of code can be protected by locks, preventing interference from concurrent threads, but also cutting down opportunities for parallelism. The past two decades have produced a wealth of data structures that instead tolerate (or even embrace) concurrency for the sake of scaling with available parallelism. But despite the importance of these "scalable" (e.g., lock-free) data structures, many basic questions about them have remained unanswered. My work has focused on several such questions; this talk will cover two of them. First, what are the fundamental abstractions for scalable concurrency? I will present Reagents, the first high-level programming model for building lock-free data structures. Reagents solve a key problem for concurrency libraries: they allow library clients to combine data structure operations while retaining strong scalability and atomicity properties, and without requiring the clients to understand the underlying algorithms. Second, how can we reason about scalable concurrency under realistic assumptions about shared memory? The verification literature almost universally assumes that threads interact through a "strong" (sequentially consistent) shared memory. But modern compilers and CPUs provide much weaker guarantees that cause many previously verified algorithms to fail. I will present GPS, the first program logic to provide a full suite of modular verification techniques under a weak memory model. Aaron Turon is a postdoctoral researcher in the Foundations of Programming group at the Max Planck Institute for Software Systems (MPI-SWS), supervised by Derek Dreyer. He received his Ph.D. from Northeastern University in May 2013 under the supervision of Mitchell Wand, and his BA with Honors at the University of Chicago in 2007. Aaron's interests span programming languages and software verification, but his primary focus for the last several years has been the design and verification of high-performance concurrency libraries.

1 1

Colloquium Speaker John Duchi Wed. Feb. 19, 4:30pm
by Nicole E. Wagenblast 19 Feb '14

19 Feb '14

Computation, Communication, and Privacy Constraints on Learning John Duchi , University of California, Berkeley Wednesday, February 19, 4:30pm Computer Science 105 How can we maximally leverage available resources--such as computation, communication, multi-processors, or even privacy--when performing machine learning? In this talk, I will suggest statistical risk (a rigorous notion of the accuracy of learning procedures) as a way to incorporate such criteria in a framework for development of algorithms. In particular, we follow a two-pronged approach, where we (1) study the fundamental difficulties of problems, bringing in tools from optimization, information theory, and statistical minimax theory, and (2) develop algorithms that optimally trade among multiple criteria for improved performance. The resulting algorithms are widely applied in industrial and academic settings, giving up to order of magnitude improvements in speed and accuracy for several problems. To illustrate the practical benefits that a focus on the tradeoffs of statistical learning procedures brings, we explore examples from computer vision, speech recognition, document classification, and web search. John is currently a PhD candidate in computer science at Berkeley, where he started in the fall of 2008. His research interests include optimization, statistics, machine learning, and computation. He works in the Statistical Artificial Intelligence Lab (SAIL) under the joint supervision of Michael Jordan and Martin Wainwright. He obtained his MA in statistics in Fall 2012, and received a BS and MS from Stanford University in computer science under the supervision of Daphne Koller. John has won several awards and fellowships, including a best student paper award at the International Conference on Machine Learning (ICML) and the NDSEG and Facebook graduate fellowships. John has also worked as a software engineer and researcher at Google Research under the supervision of Yoram Singer and Fernando Pereira.

1 2

Colloquium Speaker: Raluca Popa Tues Feb 18, 4:30pm
by Nicole E. Wagenblast 17 Feb '14

17 Feb '14

Building systems that compute on encrypted data Raluca Popa , Massachusetts Institute of Technology Tuesday, February 18, 4:30pm Computer Science 105 Theft of confidential data is prevalent. In most applications, confidential data is stored at servers. Thus, existing systems naturally try to prevent adversaries from compromising these servers. However, experience has shown that adversaries still find a way to break in and steal the data. In this talk, I will describe a new approach to protecting data confidentiality even when attackers get access to all server data: building practical systems that compute on encrypted data without access to the decryption key. In this setting, I designed and built a database system (CryptDB), a web application platform (Mylar), and two mobile systems, as well as developed new cryptographic schemes for them. I showed that these systems support a wide range of applications with low overhead. The talk will focus primarily on CryptDB and Mylar. My work has already had impact: Google uses CryptDB’s design for their new Encrypted BigQuery service, and a medical application of Boston’s Newton-Wellesley hospital is secured with Mylar. Looking forward, this approach promises to solve privacy problems in other systems, such as big data or genomics systems. Raluca Ada Popa is a PhD candidate at MIT working in security, systems, and applied cryptography. As part of her PhD work, she built practical systems that compute over encrypted data as well as designed new encryption schemes that underlie these systems. Raluca is the recipient of a Google PhD Fellowship for secure cloud computing, Johnson award for best CS Masters of Engineering thesis from MIT, and CRA Outstanding undergraduate award from the ACM. Raluca received her undergraduate degree from MIT with two BS degrees, in computer science and in mathematics.

1 2

Colloquium Speaker Tamara Broderick Wed. Feb 17 4:30pm
by Nicole E. Wagenblast 12 Feb '14

12 Feb '14

Feature allocations, paintboxes, and probability functions Tamara Broderick , University of California, Berkeley Wednesday, February 17, 4:30pm Computer Science 105 Clustering involves placing entities into mutually exclusive categories. We wish to relax the requirement of mutual exclusivity, allowing objects to belong simultaneously to multiple classes, a formulation that we refer to as "feature allocation." The first step is a theoretical one. In the case of clustering the class of probability distributions over exchangeable partitions of a dataset has been characterized (via exchangeable partition probability functions and the Kingman paintbox). These characterizations support an elegant nonparametric Bayesian framework for clustering in which the number of clusters is not assumed to be known a priori. We establish an analogous characterization for feature allocation; we define notions of "exchangeable feature probability functions" and "feature paintboxes" that lead to a Bayesian framework that does not require the number of features to be fixed a priori. The second step is a computational one. Rather than appealing to Markov chain Monte Carlo for Bayesian inference, we develop a method to transform Bayesian methods for feature allocation (and other latent structure problems) into optimization problems with objective functions analogous to K-means in the clustering setting. These yield approximations to Bayesian inference that are scalable to large inference problems. Tamara Broderick is a PhD candidate in the Department of Statistics at the University of California, Berkeley. Her research in machine learning focuses on the design and study of Bayesian nonparametric models, with particular emphasis on feature allocation as a generalization of clustering that relaxes the mutual exclusivity and exhaustivity assumptions of clustering. While at Berkeley, she has been a National Science Foundation Graduate Student Fellow and a Berkeley Fellowship recipient. She graduated with an AB in Mathematics from Princeton University in 2007---with the Phi Beta Kappa Prize for highest average GPA in her graduating class and with Highest Honors in Mathematics. She spent the next two years on a Marshall Scholarship at the University of Cambridge, where she received a Masters of Advanced Study in Mathematics for completion of Part III of the Mathematical Tripos (with Distinction) in 2008 and an MPhil by Research in Physics in 2009. She received a Masters in Computer Science from UC Berkeley in 2013.

1 2