- talks - lists.cs.princeton.edu

EE and CS Seminar - Carole-Jean Wu, Facebook AI Research/ASU, Thursday Nov 12 at 12:30pm
by Emily C. Lawrence 12 Nov '20

12 Nov '20

COMPUTER SCIENCE AND ELECTRICAL ENGINEERING SEMINAR Speaker: Carole-Jean Wu , Facebook AI Research/ASU Title: [ https://ee.princeton.edu/events/deep-learning-its-not-all-about-recognizing… | Deep Learning: It’s Not All About Recognizing Cats and Dogs ] Day: Thursday, November 12, 2020 (ET) Time: 12:30 pm Zoom Link: Please register using [ https://princeton.zoom.us/meeting/register/tJcuduGtrTguGdTbUci5h2UCDH6zISdm… | this link ] . Hosts: Margaret Martonosi & David Wentzlaff Abstract: In this seminar, we will start this seminar by examining the underinvested deep learning personalization and recommendation systems in the overall research community. The training of state-of-the-art industry-scale personalized and recommendation models consumes the highest number of compute cycles among all deep learning use cases. For AI inference, recommendation and ranking consume even higher compute cycles of 80%. What do state-of-the-art industry-scale neural personalization and recommendation models look like? I will present recent advancement on the development of deep learning recommender systems, the implications on system and architectural design and parallelism opportunities across the machine learning system stack over a variety of platforms [HPCA-2020, ISCA-2020, HPCA-2021a]. I will conclude the talk with future directions on multi-scale system design and optimization to advance the field of AI [HPCA-2019, MICRO-2020, HPCA-2021b]. Bio: Carole-Jean Wu is a Research Scientist at Facebook AI Research. Her research focus lies in the domain of computer system architecture with particular emphasis on energy- and memory-efficient systems. Her recent research has pivoted into designing systems for machine learning execution at-scale, such as for personalized recommender systems and mobile deployment. Carole-Jean chairs the MLPerf Recommendation Benchmark Advisory Board and co-chairs MLPerf Inference. Carole-Jean holds tenure from ASU. She received her M.A. and Ph.D. from Princeton and B.Sc. from Cornell. She is the recipient of the NSF CAREER Award, Facebook AI Infrastructure Mentorship Award, the IEEE Young Engineer of the Year Award, the Science Foundation Arizona Bisgrove Early Career Scholarship, and the Intel PhD Fellowship, among a number of Best Paper awards. She is a senior member of both ACM and IEEE. This seminar is supported by CS and EE Korhammer Lecture Series Funds

1 1

Tuesday 12:30: CITP seminar, Vanessa Teague on 2 problems in election integrity
by Andrew Appel 09 Nov '20

09 Nov '20

CITP seminar, Tuesday November 10 at 12:30 ( [ https://princeton.zoom.us/j/94881769712 | Zoom link ] ) Some Election Integrity Problems are Surprisingly Easy to Solve and Others are Very Very Hard by Vanessa Teague This talk will compare two recent advances in the understanding of election integrity. The Scytl/SwissPost e-voting solution was intended to provide complete verifiability for Swiss government elections. The talk will show failures in both individual verifiability and universal verifiability, based on mistaken implementations of cryptographic components. These failures allow for the construction of “proofs” of an accurate election outcome that pass verification though the votes have been manipulated. This shows that even well-regulated Internet voting systems that seem to provide some evidence of an accurate election outcome are still brittle in their assumptions and incentives, to the extent of making large-scale undetectable fraud relatively easy. Every time internet voting is reconsidered, we learn about a new aspect that makes it harder than we thought. On a more optimistic note, the talk will describe a recent pilot of Risk-Limiting Audits of Instant Runoff Votes (IRV) for the San Francisco DA’s race. Any election might be too close (or any count too error-prone) for confidence to be gained from a small sample, but IRV was once considered too complex and unstable for rigorous audits even when the margin was large. It will be shown that in practice that belief was mistaken: most real IRV elections can be analyzed and reduced to a series of well-defined assertions that can be tested efficiently! Joint work with Michelle Blom, Andrew Conway, Thomas Haines, Sarah Jamie-Lewis, Olivier Pereira, Philip Stark and Peter Stuckey. Bio: Vanessa Teague is the CEO of Thinking Cybersecurity and an associate professor (Adj.) in the Research School of Computer Science at the Australian National University. Her research focuses primarily on cryptographic methods for achieving security and privacy, particularly for issues of public interest such as election integrity and the protection of government data. She was part of the team (with Chris Culnane and Ben Rubinstein) who discovered the easy re-identification of doctors and patients in the Medicare/PBS open dataset released by the Australian Department of Health. She has co-designed numerous protocols for improved election integrity in e-voting systems, and co-discovered serious weaknesses in the cryptography of deployed e-voting systems in New South Wales, Western Australia and Switzerland. She lives and works on Wurundjeri land in Southeastern Australia (near Melbourne).

1 0

Monday, November 9, 2020 at 4:30 pm - Virtual EE Seminar - Kaushik Chowdhury, Northeastern University
by Emily C. Lawrence 06 Nov '20

06 Nov '20

ELECTRICAL ENGINEERING SEMINAR Speaker: Kaushik Chowdhury , Northeastern University Title: Deep Convolutional Neural Networks for Device Identification Day: Monday, November 9, 2020 Time: 4:30 pm Zoom Link: Meeting link below Host: Prof. Kaushik Sengupta Abstract: Network densification is poised to enable the massive throughout jump expected in the era of 5G and beyond. In the first part of the talk, we identify the challenges of verifying identify of a particular emitter in a large pool of similar devices based on unique distortions in the signal, or ‘RF fingerprints’, as it passes through a given transmitter chain. We show how deep convolutional neural networks can uniquely identify a radio in a large signal dataset composed of over a hundred WiFi radios with accuracy close to 99%. For this, we use tools from machine learning, namely, data augmentation, attention networks and deep architectures that have proven to be successful in image processing and modify these methods to work in the RF-domain. In the second part of the talk, we show how intentional injection of distortions and carefully crafted FIR filters applied to the transmitter-side can help in enhanced classification. Finally, we discuss how to detect new devices not previously seen during training using observed statistical patterns. We conclude by showing a glimpse of other applications of RF fingerprinting, like 5G waveform detection in large-scale experimental platforms and identifying a specific UAV in a swarm. Bio: Kaushik Chowdhury is Professor and Faculty Fellow in the ECE department and Associate Director at the Institute for the Wireless IoT at Northeastern University, Boston. He was awarded the Presidential Early Career Award for Scientists and Engineers (PECASE) in 2017, the DARPA Young Faculty Award in 2017, the Office of Naval Research Director of Research Early Career Award in 2016, and the NSF CAREER award in 2015. He has received best paper awards at several conferences that include, Infocom, Globecom, ICC (3x), SenSys, ICNC, and DySpan. He is presently a co-director of the Platforms for Advanced Wireless Research (PAWR) project office and the Colosseum RF emulator. His current research interests span applied machine learning to wireless systems, networked robotics, wireless charging and at-scale experimentation for emerging 5G and beyond networks. Join Zoom Meeting : [ https://princeton.zoom.us/j/99499632211?pwd=Qm4wWFFWSUFyMjJVQ3lseFh1TGNyUT09 | https://princeton.zoom.us/j/99499632211?pwd=Qm4wWFFWSUFyMjJVQ3lseFh1TGNyUT09 ]

1 0

Engage 2020 sessions featuring SEAS dean and faculty
by Emily C. Lawrence 04 Nov '20

04 Nov '20

Engage 2020, Princeton’s first innovation and entrepreneurship conference, runs this week Nov. 4-6. There are over 50 sessions aimed at helping faculty, staff and postdocs, graduate students and undergraduates learn more about building the impact of their research through innovation and entrepreneurship. Here are just a small selection of sessions. A single registration gives you access to all three days of programming: Register for [ https://princetonengage2020.pathable.co/ | Engage 2020 ] . Engage 2020 is free and open to all, including people not affiliated with Princeton. Wednesday, November 4 Educator-Entrepreneur: A Personal Journey 12:00 p.m. to 1:00 p.m. Speaker: Andrea Goldsmith and Naveen Verma, Princeton University Series: Engage 2020 Thursday, November 5 Dean for Research Award for Distinguished Innovation: Honorary Lecture by Professor Robert Prud'homme 12:00 p.m. to 1:00 p.m. Speaker: Robert Prud’homme, Princeton University Series: Engage 2020 The Future of Collaboration and Inclusive Innovation: Microsoft + Princeton 2:00 p.m. to 3:30 p.m. Speaker: Jen Rexford, '91, Princeton University; Brad Smith, ’81 and President of Microsoft Series: Engage 2020 Celebrate Princeton Innovation: Technology Showcase 3:30 p.m. to 6 p.m. Speakers: Pablo Debenedetti, Rodney Priestley, Daniel Powell, co-founder of a startup based on the work of Warren Powell, Craig Arnold, Dalton Conley, Sujit Datta, Phil Efthimion from PPPL, Ning Lin, David MacMillan, Naveen Verma, Xiao Fan on behalf of Nieng Yan Series: Engage 2020 Friday, November 6 Princeton's Pandemic Research and Solutions: Spotlight on Several Princeton-Funded COVID-19 Research Projects 2:00 p.m. to 2:55 p.m. Speakers: Cliff Brangwynne, Kyle Jamieson, Natalie Cox, Princeton University; Nishith Prakash, Harvard University Series: Engage 2020 Overcome the Funding Hurdle: Funding My Clean Energy Startup 3:00 p.m. to 3:55 p.m. Speakers: Angelo Campus, ’16; Andrew Hsieh, ’10, *14; Stephanie Spiers, *14; Elizabeth Washburn Surti, ‘08 Series : Engage 2020 Engage 2020 Faculty and Alumni Startup Showcase 5:00 p.m. to 7:00 p.m. Speakers: Ed Felten and Harry Kalodner, *16, Offchain Labs, Inc.; Marcus Hultmark and Yuyang Fan, *17; Mark Esposito, *17; Princeton University; and other company co-founders Series : Engage 2020 Featured Tracks: Core Facilities Showcase (Wed 9:30 a.m. – 4:30 p.m.): Now in its second year, the Showcase sessions and virtual exhibit booths will highlight research services and equipment offered by academic institutions from around the Garden State. Featuring Craig Arnold, Nan Yao, Jared Toettcher, Nieng Yan and more Data Science in Oncology and Immunology (Thurs 10 a.m. – 4 p.m.): In this in-depth workshop, industry partners and academic researchers will explore areas of mutual interest and identify potential collaboration in the areas of biomedical data science, computational oncology and computational immunology. Featuring Ben Raphael, Barbara Engelhardt, Curtis Callan, Olga Troyanskaya, Mona Singh Entrepreneurship Track (Fri 9 a.m. – 7 p.m.): Webinars, skill-building workshops and Princeton-related startup showcases will offer entrepreneurs and those who work with them an abundance of thought-provoking opportunities. Catherine Zandonella | Communications Manager Office of the Dean for Research | Princeton University 91 Prospect Ave. | Princeton | NJ 08540 [ mailto:czandone@princeton.edu | czandone(a)princeton.edu ] | 609-258-0541 [ https://research.princeton.edu/ | research.princeton.edu ] [ http://www.facebook.com/princetonuniversityresearch | facebook ] [ http://www.twitter.com/princetonideas | @princetonideas ] [ http://discovery.princeton.edu/ | Discovery: Research at Princeton ]

1 0

TODAY at 4:30 PM - Virtual IEEE SSCS Distinguished Lecture Colloquium - Tuesday, November 3, 2020
by Emily C. Lawrence 03 Nov '20

03 Nov '20

TODAY - Sent on behalf of Prof. Kaushik Sengupta IEEE Princeton/Central Jersey Section Distinguished Lecture Colloquium, November 2020 Speaker: Prof. Patrick Mercier , University of California, San Diego Talk Title : Low-Power Communication Circuits and Technologies for Emerging Internet-of-Things Applications Date: Tuesday, November 3, 2020 Time: 4:30 – 5:30 p.m. Zoom link: Meeting link below Abstract: Emerging Internet-of-Things (IoT) devices for use in smart homes, wearable systems, industrial monitoring, smart cities, and beyond all require robust yet low-power wireless communications. Unfortunately, most current wireless standards do not intrinsically support low-power operation due to strict requirements on modulation formats, data rates, linearity, packet overheads, and so on. These restrictions impose minimum power consumption requirements for cellular standards (e.g., GSM, LTE, and 5G) and WiFi, but also surprisingly limit the ability of supposedly low-power standards (e.g., Bluetooth Low Energy and Narrowband-IoT) from reaching new application-enabling power levels. This presentation will outline the major challenges facing power reduction in modern wireless systems, and will describe several possible solutions to these challenges. Specifically, we will explore the use of wake-up receivers as a means to reduce the power overhead of between-node synchronization, including a look at a recent design operating at 9GHz. Then, we will discuss a variety of alternative communication schemes that can help to reduce the power of communication in WiFi and body-area-network systems by >1,000x through use of WiFi-compliant backscatter communication and magnetic human body communication systems, respectively. We will then introduce new circuit design techniques based on a topologically-derived “digitally-replaced analog” paradigm towards the implementation of efficient Li-ion-battery-connected RF power amplifiers for high-performance wireless systems. Bio : Patrick Mercier is an Associate Professor of Electrical and Computer Engineering and co-founder/co-director of the Center for Wearable Sensors at UC San Diego. He received his B.Sc. degree from the University of Alberta, Canada, in 2006, and the S.M. and Ph.D. degrees from MIT in 2008 and 2012, respectively. Prof. Mercier has received numerous awards, including the San Diego Engineering Council Outstanding Engineer Award in 2020, a National Academy of Engineering Frontiers of Engineering Speaker in 2019, the NSF CAREER Award in 2018, the Biocom Catalyst Award in 2017, the UCSD Academic Senate Distinguished Teaching Award in 2016, the DARPA Young Faculty Award in 2015, the Beckman Young Investigator Award in 2015, The Hellman Fellowship Award in 2014, the International Solid-State Circuits Conference (ISSCC) Jack Kilby Award in 2010, amongst others. He has published over 140 peer-reviewed papers in venues such as Nature Biotechnology, Nature Communications, ISSCC (22 papers), Advanced Science, and others. He is an Associate Editor of the IEEE Transactions on Biomedical Circuits and Systems and the IEEE Solid-State Circuits Letters, is a member of the ISSCC, CICC, and VLSI Technical Program Committees, and has co-edited three books: High-Density Integrated Electrocortical Neural Interfaces (Elsevier Academic Press, 2019), Power Management Integrated Circuits (CRC Press, 2016), and Ultra-Low-Power Short-Range Radios (Springer, 2015). His research interests include the design of energy-efficient mixed-signal systems, RF circuits, power converters, and sensor interfaces for wearable, medical, and mobile applications. Join Zoom Meeting : [ https://princeton.zoom.us/j/99499632211?pwd=Qm4wWFFWSUFyMjJVQ3lseFh1TGNyUT09 | https://princeton.zoom.us/j/99499632211?pwd=Qm4wWFFWSUFyMjJVQ3lseFh1TGNyUT09 ] IEEE Princeton/Central Jersey Section: [ http://ewh.ieee.org/r1/princeton-centraljersey/ | http://ewh.ieee.org/r1/princeton-centraljersey/ ] Contact Info: Shubha Bommalingaiahnapallya, [ mailto:shubha.bpallya@gmail.com | shubha.bpallya(a)gmail.com ] , Kaushik Sengupta, [ mailto:kaushiks@princeton.edu | kaushiks(a)princeton.edu ] or Hooman Saeidi , [ mailto:hsaeidi@princeton.edu | hsaeidi(a)princeton.edu ]

1 0

Zoe Paraskevopoulou will present her FPO "Verified Optimizations for Functional Languages" on Friday, 10/30/2020 at 3pm via Zoom.
by Nicki Mahler 23 Oct '20

23 Oct '20

Zoe Paraskevopoulou will present her FPO "Verified Optimizations for Functional Languages" on Friday, 10/30/2020 at 3pm via Zoom. Zoom link: [ https://princeton.zoom.us/j/5655804837 | https://princeton.zoom.us/j/5655804837 ] The members of her committee are as follows: Andrew Appel (Adviser); Readers: Davis Walker and Amal Ahmed (Northeastern University); Examiners: Zachary Kincaid, Aarti Gupta, Davis Walker, Andrew Appel, Amal Ahmed (Northeastern University) A copy of her thesis is available upon request. Please email ngotsis(a)cs.princeton if you would like a copy of the thesis. Everyone is invited to attend her talk. Abstract follows below: Coq is one of the most widely adopted proof development systems. It allows programmers to write purely functional programs and verify them against specifications with machine-checked proofs. After verification, one can use Coq’s extraction plugin to obtain a certified program (in OCaml, Haskell, or Scheme) that can be compiled and executed. However, bugs in either the extraction function or the compiler of the extraction language, can render source level verification useless. A verified compiler is a compiler whose output provably preserves the semantics of the source language. CertiCoq is a currently developed verified compiler for Gallina, Coq’s specification language. With CertiCoq, one could obtain an executable program that has is guaranteed to have the same behavior as the source program, bridging the gap between the formally verified source program and the executable. In this thesis, I present the implementation and verification of CertiCoq’s optimizing middle-end pipeline. CertiCoq’s middle end consists of seven di↵erent transformations and is responsible for eciently compiling an untyped purely functional intermediate language to a first-order subset of this language, which can be readily compiled to a first-order, low-level intermediate language. CertiCoq’s middle-end pipeline performs crucial optimizations for functional languages including closure conversion, uncurrying, shrink-reduction and inlining. It advances the state of the art of verified optimizing compilers for functional languages by implementing ecient closure-allocation strategies. For proving CertiCoq correct, I develop a framework based on the technique of logical relations, making novel technical contributions. I extend the logical relation with notions of precondition and postcondition showing how it can be used to reason about intensional properties of programs simultaneously with extensional properties. I demonstrate how this allows reasoning about divergence preservation, which is not supported by traditional logical relations. I develop a novel, lightweight technique that allows logical-relation proofs to be composed, enabling verified separate compilation of programs compiled with CertiCoq using di↵erent sets of optimizations. Lastly, I use the framework to prove that CertiCoq’s closure conversion is not only functionally correct but also safe for time and space, meaning that it is guaranteed to preserve the asymptotic time and space complexity of the source program.

1 0

Malte Moeser will present his Pre- FPO on Wednesday, October 28, 2020 via Zoom
by Nicki Mahler 21 Oct '20

21 Oct '20

Malte Moeser will present his Pre- FPO on Wednesday, October 28, 2020 via Zoom. Zoom link: [ https://princeton.zoom.us/j/92656073526 | https://princeton.zoom.us/j/92656073526 ] The members of his committee are as follows: Arvind Narayanan (advisor and examiner) Jonathan Mayer (examiner) Nicolas Christin (Carnegie Mellon University, examiner) Matt Weinberg (reader) Rainer Böhme (University of Innsbruck, reader) Link to abstract below: [ https://docs.google.com/document/d/1D6mJuNFLJRdfVIcr94jht1Vsk4kkM4bhB0PrXvD… | https://docs.google.com/document/d/1D6mJuNFLJRdfVIcr94jht1Vsk4kkM4bhB0PrXvD… ]

1 0

Laura Roberts will present her FPO, "Internet Measurement for the Prevention and Detection of Internet Surveillance" on Wednesday, 10/28/2020 at 2pm via Zoom.
by Nicki Mahler 21 Oct '20

21 Oct '20

Laura Roberts will present her FPO, "Internet Measurement for the Prevention and Detection of Internet Surveillance" on Wednesday, 10/28/2020 at 2pm via Zoom. Zoom link: [ https://princeton.zoom.us/j/96919654569 | https://princeton.zoom.us/j/96919654569 ] The members of her committee are as follows: Ed Felten (adviser); Readers: Prateek Mittal and Jonathan Mayer; Examiners: Jennifer Rexford, Arvind Narayanan, and Ed Felten. A copy of her thesis is available upon request. Please email ngotsis(a)cs.princeton if you would like a copy of the thesis. Everyone is invited to attend her talk. Abstract follows below: Surveillance is a tactic of societal control: Observation of people’s behavior allows the observer to manipulate and steer people’s future behavior in a way that serves the observer’s interests. With new technologies come new, easier ways of surveilling people. This dissertation concerns itself with Internet technologies in particular and presents how Internet measurement can be used to prevent and detect Internet surveillance. "Internet surveillance" is the observation of people’s online activities. "Internet measurement" comprises technical means of measuring interesting characteristics of the Internet (e.g. packet round-trip time, protocol usage, etc.) in order to gain better understanding of the Internet and improve its performance, security, etc. So in other words, this dissertation presents how various Internet measurement techniques can be used for preventing and detecting the observation of people’s online activities. In order to achieve the goal of preventing surveillance, we use Internet measurement techniques to strengthen the Tor anonymity network (a system that seeks to thwart Internet surveillance and censorship) by exposing vulnerabilities and suggesting possible solutions. In order to achieve the goal of detecting surveillance, we develop a proof of concept system that uses Internet measurement techniques to catch surveillants.

1 0

Friday, Oct 23 talk by Whitney Quesenbery, Center for Civic Design
by Emily C. Lawrence 20 Oct '20

20 Oct '20

The Electoral Innovation Lab at Princeton University presents a Friday Forum: Election Reform as a Design Challenge, featuring Whitney Quesenbery, Executive Director of the Center for Civic Design. Friday, October 23, at 1:30pm EDT via Zoom: [ https://princeton.zoom.us/j/273691213 | https://princeton.zoom.us/j/273691213 ] Please RSVP to [ mailto:coopersmith@princeton.edu | coopersmith(a)princeton.edu ] About the speaker: Whitney Quesenbery works with election offices around the country to improve the voter experience and introduce best practices for election design. Her work in civic design began with her appointment to the US Election Assistance Commission (EAC) federal advisory committee writing the Voluntary Voting System Guidelines, including the first usability requirements for voting systems. She is a co-author of the Brennan Center for Justice’s seminal publication, Better Ballots , and an expert in their Ballot Design project. As the coordinator for the EAC-funded Accessible Voting Technology Initiative, she has managed 15 grants on voting for people with disabilities, including the development of the Anywhere Ballot. Among her many professional commitments, Quesenbery is a member of the Election Modernization Advisory Committee for the state of Michigan.

1 0

Ashwini Raina will present his General Exam, "Read-Aware LSM for Disaggregated Heterogeneous Storage" on Monday, October 26, 2020 at 8:30am via Zoom.
by Nicki Mahler 19 Oct '20

19 Oct '20

Ashwini Raina will present his General Exam "Read-Aware LSM for Disaggregated Heterogeneous Storage" on Monday, October 26, 2020 at 8:30am via Zoom. Zoom link: [ https://princeton.zoom.us/j/94253551655 | https://princeton.zoom.us/j/94253551655 ] The members of his committee are as follows: Michael Freedman, Kyle Jamieson, and Wyatt Lloyd Link to Abstract and Reading List: [ https://docs.google.com/document/d/1tKGtWXGrKuzJbkGBw3Ite6t4gvUBjo0S5i9-Q8c… | https://docs.google.com/document/d/1tKGtWXGrKuzJbkGBw3Ite6t4gvUBjo0S5i9-Q8c… ] [ https://docs.google.com/document/d/1i_AaAf-P7_LxNugxrgNeRjaki6NVn27FmoXvN5W… | https://docs.google.com/document/d/1i_AaAf-P7_LxNugxrgNeRjaki6NVn27FmoXvN5W… ] Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

1 0