- talks - lists.cs.princeton.edu

REMINDER: Speaker-Sreepathi Pai, tomorrow at 12:30
by Emily C. Lawrence 11 Oct '18

11 Oct '18

Speaker Sreepathi Pai from University of Rochester Friday, October 12, 2018 - 12:30pm Computer Science - Room 105 Host: Tyler Sorensen and Margaret Martonosi “High-performance Graph Processing on GPUs” Abstract: Back in 2013, implementing high-performance graph algorithms for GPUs required manual coding in CUDA, a slow and difficult process. To significantly lower the level of effort required, we created IrGL , a language and compiler specifically for generating high-performance graph algorithm implementations for GPUs. Powered by three throughput optimizations, the IrGL -generated code outperformed nearly all handwritten graph algorithms achieving speedups of up to 6x. Freed from the drudgery of writing low-level code, IrGL has allowed us to look at a number of problems revolving around graphs. We've used the high-performance implementations to identify key memory system bottlenecks that limit performance on current GPU architectures. We've also translated graph database queries to IrGL and executed them on GPUs. In the course of extending this to the general problem of subgraph isomorphism (a key primitive in graph databases), we were named GraphChallenge 2017 champions for our implementation of the triangle-counting and k-truss problems. Along the way, we also built Groute , a runtime for asynchronous multi-GPU graph analytics that has achieved order-of-magnitude improvements over existing synchronous implementations. We sped up exhaustive testing of software by traversing graphs that were too big to materialize in memory. Recently, we have also used IrGL's ability to generate hundreds of variants of the same graph algorithm to explore correctness and performance portability issues on GPUs. Many interesting questions still remain unexplored, however, and I will summarize our current efforts in this area. [Joint work with Keshav Pingali , Tal Ben-Nun, Michael Sutton, M. Amber Hassaan , Chad Voegele , Yi-Shan Lu, Ahmet Celik , Milos Gligoric , Sarfraz Khurshid , Tyler Sorensen and Alastair Donaldson] Bio: Sreepathi Pai is an Assistant Professor of Computer Science at the University of Rochester. His research interests are in compilers, programming languages and implementation, performance models and computer architecture. His most recent research has revolved around the IrGL compiler that produces high-performance GPU code for graph algorithms. He earned his PhD at the Indian Institute of Science and his B.E. in Computer Engineering at the University of Mumbai. Prior to joining the Department of Computer Science at Rochester, he was a Postdoctoral Fellow at the University of Texas at Austin.

1 0

Colloquium Speaker-Andy Myers '96, TODAY at 12:30
by Emily C. Lawrence 09 Oct '18

09 Oct '18

Colloquium Speaker Andy Myers ’96 from Jane Street Tuesday, October 9, 2018 - 12:30pm Computer Science - Room 105 Host: Jennifer Rexford & David Walker *Lunch available for talk attendees at 12:00 pm * “Trading Networks” Abstract: The talk will discuss some of the unique ways in which trading firms use (and misuse) various bits of technology (switches, layer-1 devices, microwaves, FPGAs), to build networking infrastructure that achieves their goals. Bio: Andy Myers has been building low latency trading systems at Jane Street since 2014. Previously he was a Systems Architect at GETCO. He received his A.B. from Princeton in 1996 and his MS from Carnegie Mellon in 1998.

1 0

Speaker-Sreepathi Pai, Friday, October 12
by Emily C. Lawrence 08 Oct '18

08 Oct '18

Speaker Sreepathi Pai from University of Rochester Friday, October 12, 2018 - 12:30pm Computer Science - Room 105 Host: Margaret Martonosi “High-performance Graph Processing on GPUs” Abstract: Back in 2013, implementing high-performance graph algorithms for GPUs required manual coding in CUDA, a slow and difficult process. To significantly lower the level of effort required, we created IrGL , a language and compiler specifically for generating high-performance graph algorithm implementations for GPUs. Powered by three throughput optimizations, the IrGL -generated code outperformed nearly all handwritten graph algorithms achieving speedups of up to 6x. Freed from the drudgery of writing low-level code, IrGL has allowed us to look at a number of problems revolving around graphs. We've used the high-performance implementations to identify key memory system bottlenecks that limit performance on current GPU architectures. We've also translated graph database queries to IrGL and executed them on GPUs. In the course of extending this to the general problem of subgraph isomorphism (a key primitive in graph databases), we were named GraphChallenge 2017 champions for our implementation of the triangle-counting and k-truss problems. Along the way, we also built Groute , a runtime for asynchronous multi-GPU graph analytics that has achieved order-of-magnitude improvements over existing synchronous implementations. We sped up exhaustive testing of software by traversing graphs that were too big to materialize in memory. Recently, we have also used IrGL's ability to generate hundreds of variants of the same graph algorithm to explore correctness and performance portability issues on GPUs. Many interesting questions still remain unexplored, however, and I will summarize our current efforts in this area. [Joint work with Keshav Pingali , Tal Ben-Nun, Michael Sutton, M. Amber Hassaan , Chad Voegele , Yi-Shan Lu, Ahmet Celik , Milos Gligoric , Sarfraz Khurshid , Tyler Sorensen and Alastair Donaldson] Bio: Sreepathi Pai is an Assistant Professor of Computer Science at the University of Rochester. His research interests are in compilers, programming languages and implementation, performance models and computer architecture. His most recent research has revolved around the IrGL compiler that produces high-performance GPU code for graph algorithms. He earned his PhD at the Indian Institute of Science and his B.E. in Computer Engineering at the University of Mumbai. Prior to joining the Department of Computer Science at Rochester, he was a Postdoctoral Fellow at the University of Texas at Austin.

1 0

Reminder: Colloquium Speaker-Andy Myers '96, tomorrow at 12:30
by Emily C. Lawrence 08 Oct '18

08 Oct '18

Colloquium Speaker Andy Myers ’96 from Jane Street Tuesday, October 9, 2018 - 12:30pm Computer Science - Room 105 Host: Jennifer Rexford & David Walker *Lunch available for talk attendees at 12:00pm* “Trading Networks” Abstract: The talk will discuss some of the unique ways in which trading firms use (and misuse) various bits of technology (switches, layer-1 devices, microwaves, FPGAs), to build networking infrastructure that achieves their goals. Bio: Andy Myers has been building low latency trading systems at Jane Street since 2014. Previously he was a Systems Architect at GETCO. He received his A.B. from Princeton in 1996 and his MS from Carnegie Mellon in 1998.

1 0

Colloquium Speaker-Andy Myers '96, Tuesday, October 9
by Emily Lawrence 02 Oct '18

02 Oct '18

Colloquium Speaker Andy Myers '96 from Jane Street Tuesday, October 9, 2018 - 12:30pm Computer Science - Room 105 Host: Jennifer Rexford & David Walker "Trading Networks" Abstract: The talk will discuss some of the unique ways in which trading firms use (and misuse) various bits of technology (switches, layer-1 devices, microwaves, FPGAs), to build networking infrastructure that achieves their goals. Bio: Andy Myers has been building low latency trading systems at Jane Street since 2014. Previously he was a Systems Architect at GETCO. He received his A.B. from Princeton in 1996 and his MS from Carnegie Mellon in 1998.

1 0

TODAY 10/1 at 4:30pm in B205 - Chandra Nair, Chinese University of Hong Kong
by Emily Lawrence 01 Oct '18

01 Oct '18

EE SEMINAR SERIES Speaker: Chandra Nair, Chinese University of Hong Kong Title: Non-Convex Optimization And Multiuser Information Theory Day: Monday, October 1, 2018 Time: 4:30 pm Room: B205 Engineering Quadrangle Host: Prof. Peter Ramadge Abstract: Capacity regions in multiuser information theory were traditionally established using an achievability argument and converses to corresponding rate regions. Recently, however, capacity regions have been established, for non-trivial and important classes of channels (such as additive Gaussian noise models and others), by optimizing functionals on probability spaces generated from inner or outer bounds and showing that the bounds match for these channels. Optimization ideas have also been used to determine if certain achievable regions proposed in the literature are optimal or sub-optimal in general. The study of these non-convex optimization problems required developing new insights and techniques. In this talk, I will outline the ideas involved in obtaining the results, as well as illustrate the relationship of these problems to hypercontractivity and related notions studied primarily outside information theory. I will present some unifying observations across the family of optimization problems that we have managed to solve, and state some elementary instances of similar problems whose solutions would be of immense interest. Bio: Chandra Nair is an Associate Professor with the Information Engineering department at The Chinese University of Hong Kong. His research interests and contributions have been in developing ideas, tools, and techniques to tackle families of combinatorial and non-convex optimization problems arising primarily in the information sciences. His recent research focus has been on studying the optimality of certain inner and outer bounds to capacity regions for fundamental problems in multiuser information theory. He, along with his doctoral student Yanlin Geng, received the 2016 Information Theory Society paper award for developing a novel way to establish the optimality of Gaussian distributions. A proof of the Parisi and Coppersmith-Sorkin conjectures in the Random Assignment Problem constituted his doctoral dissertation; and he resolved some conjectures related to Random Energy model approximation of the Number Partition Problem during his post-doctoral years. Chandra Nair got his Bachelor's degree, B.Tech(EE), from IIT Madras (India) where he was the Philips (India) and Siemens (India) award winner for the best academic performance. Subsequently he was a Stanford graduate fellow (00-04) and a Microsoft graduate fellow (04-05) during his graduate studies at the EE department of Stanford university. Later, he became a post-doctoral researcher (05-07) with the theory group at Microsoft Research, Redmond. He has been a faculty member of the information engineering department at The Chinese university of Hong Kong since Fall 2007. He was an associate editor for the IEEE Transactions on Information Theory (2014-2016) and is currently a distinguished lecturer of the IEEE Information theory society. He is a fellow of the IEEE. He serves as the Programme Director of the undergraduate program on Mathematics and Information Engineering and the Director of the Institute of Theoretical Computer Science and Communications.

1 0

Save the date for Princeton Research Day - May 9, 2019
by Emily Lawrence 01 Oct '18

01 Oct '18

Save the Date for Princeton Research Day! Posters . Talks . Performances . Art Exhibitions . Digital Presentations Thursday, May 9, 2019 Frist Campus Center <http://researchday.princeton.edu/> researchday.princeton.edu #PRD19 We encourage all academic departments and centers at Princeton to reserve May 9, 2019, for Princeton Research Day, and to support participation by your undergraduates, graduate students, postdoctoral researchers and fellows, and other early career researchers. Princeton Research Day is a celebration of the research and creative endeavors by our undergraduates, graduate students, postdoctoral researchers and fellows, and other early career researchers. The fourth-annual campus-wide event will feature poster presentations, exhibits, performances, selected talks and an award ceremony for outstanding contributions. The all-day event is free and open to the public. Application will open Dec. 3, 2018. Dates to remember: Dec. 3: Call for applications to present at PRD19 Feb. 18: Priority application deadline April 1: Presenters informed of presentation format and given access to edit presentation summaries April 22: Presenter RSVP deadline; last day to edit presentation summaries April 26: PRD19 program becomes available to public May 9: Princeton Research Day More details are available at researchday.princeton.edu <http://researchday.princeton.edu/> . Questions? Contact researchday@princeton <https://owa.princeton.edu/owa/redir.aspx?C=RLzvZJLd5qsfA35tEu-RZUskcIzw1Or- xtqM97sHprDuM74sownUCA..&URL=mailto%3aresearchday%40princeton.edu> .edu <https://owa.princeton.edu/owa/redir.aspx?C=YX0gtaHjrc-9tcOtKgiNyz3ODOIM6cxn odqEzb_K5mX_Wr4sownUCA..&URL=mailto%3aresearchday%40princeton.edu> . Princeton Research Day is a collaborative initiative between the offices of the Dean of the College, the Dean of the Faculty, the Dean of the Graduate School and the Dean for Research, with support from the Office of the Provost. Alice Seneres, Ph.D. Assistant Dean for Academic Affairs Office of the Dean of the Faculty Princeton University 306A Nassau Hall (609) 258-4793 aseneres(a)princeton.edu <mailto:aseneres@princeton.edu>

1 0

Seminar-Dr. Shijia Pan, Thursday, September 27
by Emily Lawrence 26 Sep '18

26 Sep '18

Seminar Speaker Dr. Shijia Pan from Carnegie Mellon University Thursday, September 27, 2018 - 12:30pm Computer Science - Room 402 Host: Margaret Martonosi "Indoor Human Information Acquisition from Physical Vibrations" Abstract: The number of everyday smart devices (such as smart TV, Samsung SmartThings, Nest, Google Home) is projected to grow to the billions in the coming decade. The Cyber-Physical Systems or Internet of Things systems that consist of these devices are used to obtain human information for various smart building applications. Different sensing approaches have been explored, including vision-, sound-, RF-, mobile-, and load-based methods, to obtain various indoor human information. >From the system perspective, general problems faced by these existing technologies are their sensing requirements (e.g., line-of-sight, high deployment density, carrying a device) and intrusiveness (e.g., privacy concerns). My research focuses on non-intrusive indoor human information acquisition through ambient structural vibration, which is referred to as 'structures as sensors'. People's interaction with structures in the ambient environment (e.g., floor, table, door) induces those structures to vibrate. By capturing and analyzing the vibration response of structures, we can indirectly infer information about the people and their actions that cause it. However, challenges remain. Due to the complexity of the physical world (in this case, both structures and people), sensing data distributions can change significantly under different sensing conditions. Therefore, from the data perspective, accurate information learning through a pure data-driven approach requires a large amount of labeled data, which is costly and difficult if not impossible to obtain in real-world sensing applications. My research addresses these challenges by combining physical knowledge and data-driven approaches. Specifically, my system can robustly learn human information from limited labeled data distributions by iteratively expanding the labeled dataset. With insights into the relationship between changes of sensing data distributions and measurable physical attributes, the iterative algorithm guides the expansion order by measured physical attributes to ensure a high learning accuracy in each iteration. Bio: Dr. Shijia Pan is a postdoctoral researcher at Carnegie Mellon University. She received her Bachelor's degree in Computer Science and Technology from University of Science and Technology of China and her Ph.D. degree in Electrical and Computer Engineering at Carnegie Mellon University. Her research interests include cyber-physical systems, Internet-of-Things (IoT), and ubiquitous computing. She worked in multiple disciplines and focused on indoor human information acquisition through ambient sensing. She has published in both top-tier Computer Science ACM/IEEE conferences (IPSN, UbiComp) and high-impact Civil Engineering journals (Journal of Sound and Vibration, Frontiers Built Environment). She is the recipient of numerous awards and fellowships, including Rising Stars in EECS, Nick G. Vlahakis Graduate Fellowship, Google Anita Borg Scholarship, Best Poster Awards (SenSys, IPSN), Best Demo Award (Ubicomp), Best Presentation Award (SenSys Doctoral Colloquium), and Audience Choice Award (BuildSys) from ACM/IEEE conferences.

2 1

Today's Department Colloquium at 4:30 pm Sherrerd Hall 101: Yuval Peres, Microsoft
by Emily Lawrence 25 Sep '18

25 Sep '18

===== Todays ORFE Department Colloquium Announcement===== DATE: Tuesday, September 25th, 2018 TIME: 4:30pm LOCATION: Sherrerd Hall 101 SPEAKER: Yuval Peres, Microsoft Research Title: Communication cost of consensus for nodes with limited memory Abstract: Motivated by applications in blockchain and sensor networks, we consider a model of n nodes trying to reach consensus on their majority bit (which a fraction p>1/2 of them possess; the rest start with the minority bit). Each node v is a finite automaton with m bits of memory (i.e., 2^m states), and a Poisson clock. When it rings, the node v can choose to communicate, and is then matched to a uniformly chosen node u, whence both v,u can update their states based on the other nodes state. Previous work has focused on minimizing the time to consensus and the probability of error; our goal is minimizing communication. We show that when m>3 logloglog(n), consensus can be reached at linear communication cost, but this is impossible if m<logloglog(n). The best algorithms we know intrinsically select a set of expert nodes that learn the majority bit, and then disseminate this knowledge. A key step is to distinguish when nodes can become aware of reaching the majority bit and stop communicating; this is impossible if their memory is too low. (Joint work with Giulia Fanti, Nina Holden and Gireeja Ranade). _____ Biography: Yuval Peres is a Principal Researcher at Microsoft Research in Redmond. He obtained his Ph.D. at the Hebrew University, Jerusalem in 1990, and later taught there and at the University of California, Berkeley. He has published more than 300 research papers and has co-authored books on Brownian motion, Markov chains, Probability on Networks, Fractals, and Game Theory. Yuval was awarded the Rollo Davidson Prize in 1995, the Loève Prize in 2001, and the David P. Robbins Prize in 2011. He was an invited speaker at the 2002 International Congress of Math. and in the 2008 European Congress, and was a plenary lecturer at the Math. Congress of the Americas in 2017. He has advised 22 Ph.D. students. He is a fellow of the American Math. Society and the Institute of Mathematical Statistics. In 2016 he was elected to the U.S. National Academy of Sciences.

1 0

Fwd: CS/EE Seminar - Monday, September 17. 2018, 11:00 am, E-Quad B327 - Liang Wang
by Jennifer Rexford 14 Sep '18

14 Sep '18

> > > Department of Computer Science and > Department of Electrical Engineering Seminar > > > Speaker: Liang Wang > University of Wisconsin - Madison > Title: Understanding Security and Privacy Issues in Cloud Applications > Date: Monday, September 17, 2018 > Time: 11:00 am > Room: E-Quad, Room B327 > Host: Prof. Jennifer Rexford and Prof. Prateek Mittal > > Abstract: An increasing amount of applications are utilizing public clouds and rely on their cloud providers to provide security and privacy protection. However, specific application design patterns might compromise the providers’ efforts on securing applications, and make the applications vulnerable. In my work, I use measurement-driven approaches to examine various types of cloud-based applications/services and investigate whether they can achieve desired security and privacy guarantees. In this talk, I will present my work on: > > (1) Decision-tree-based traffic analysis attacks against Tor meek, a cloud-based traffic obfuscator that is designed to be resistant to Internet censorship by nation states. Our attacks require a censor to maintain little state, but can reliably detect meek traffic with high true-positive rates and sufficiently low false-positive rates (0.006% - 0.02%, out of 14M TCP flows). > > (2) Side-channel attacks against search indexes of multi-tenant cloud services. The attacks can discover terms in other tenants’ private documents, and work efficiently on live cloud services such as GitHub in controlled experiments. > > (3) My most recent work on characterizing popular serverless computing platforms in terms of performance and resource scheduling. We discover several issues that could make serverless applications vulnerable to side-channel and DoS attacks in AWS Lambda and Azure Functions, and find bugs that allow customers to use resources for free in Google Cloud Functions. > > >

1 0