Quick links

Seminar

Future Vehicular Radars: Two Aspects

Date and Time
Friday, August 13, 2021 - 1:00pm to 2:00pm
Location
Zoom Webinar (off campus)
Type
Seminar
Speaker
Sumit Roy, from University of Washington
Host
Yasaman Ghasempour (ECE)

Please register here

This talk will present an overview of recent research in FUNLAB around the use of vehicular radar for advanced driver assistance systems (en route to a future vision of autonomous driving). Wideband (typically FMCW or chirp) radars are being increasingly deployed onboard vehicles as key high-resolution sensor for environmental mapping/imaging and various safety features. The talk will focus on the evolving role of radar `cognition’ in complex operating environments to address two important future challenges:

1.  Mitigating multi-access interference among Radars (e.g. dense traffic scenario)
The talk will first illustrate the impact of mutual interference on detection performance in Chirp/FMCW radars and then highlight some multi-access protocol design approaches for effective resource sharing among multiple radars.

2.  Contributions to radar vision via new radar hardware (MIMO radar) + associated advanced signal processing (Synthetic Aperture principles) as well as Convolutional Neural Network (`Radar Net’) based machine learning approach for enhanced object detection/classification in challenging circumstances.

Bio: Sumit Roy received the B. Tech. degree from the Indian Institute of Technology (Kanpur) in 1983, and the M. S. and Ph. D. degrees from the University of California (Santa Barbara), all in Electrical Engineering in 1985 and 1988 respectively, as well as an M. A. in Statistics and Applied Probability in 1988. He is currently Professor, Electrical & Comp. Engineering, appointed to a term Distinguished Professorship for Integrated Systems (2014-19); since Sep. 2020, he serves as a rotator for US Dept. of Defense’s Program Lead for Innovate Beyond 5G https://5g-to-xg.org. His research interests include fundamental design and evaluation of wireless communication and sensor network systems spanning a diversity of technologies and application areas: next-gen (5G & beyond) wireless LANs and cellular networks, heterogeneous network coexistence, spectrum sharing, software defined radio platforms, vehicular and sensor networking.  He spent 2001-03 on academic leave at Intel Wireless Technology Lab as a Senior Researcher engaged in systems architecture and standards development for ultra-wideband systems (Wireless PANs) and next generation high-speed wireless LANs. During Jan-July 2008, he was Science Foundation of Ireland’s E.T.S.   Walton   Awardee for a sabbatical at University College, Dublin and also recipient of a Royal Acad. Engineering (UK) Distinguished Visiting Fellowship during summer 2011.  He has served as IEEE Communications Society (ComSoc) Distinguished Lecturer and as Associate Editor for all the major ComSoc journals. He served 2 terms as (elected) member of Executive Committee, National Spectrum Consortium dedicated to efficient spectrum sharing between Federal and commercial networks and is the co-author of IEEE TAES 2016 Best paper award for work on Radar-Comm coexistence.  He was elevated to IEEE Fellow by Communications Society in 2007 for ``contributions to multi-user communications theory and cross-layer design of wireless networking standards”.

Putting AI on a Diet: TinyML and Efficient Deep Learning

Date and Time
Wednesday, May 19, 2021 - 10:30am to 11:30am
Location
Zoom Webinar (off campus)
Type
Seminar
Host
Kai Li and Naveen Verma

Song Han

Please register here


Today’s AI is too big. Deep neural networks demand extraordinary levels of data and computation, and therefore power, for training and inference. This severely limits the practical deployment of AI in edge devices. We aim to improve the efficiency of neural network design. First, I’ll present MCUNet [1] that brings deep learning to IoT devices. MCUNet is a framework that jointly designs the efficient neural architecture (TinyNAS) and the light-weight inference engine (TinyEngine), enabling ImageNet-scale inference on micro-controllers that have only 1MB of Flash. Next I will introduce Once-for-All Network [2], an efficient neural architecture search approach, that can elastically grow and shrink the model capacity according to the target hardware resource and latency constraints. From inference to training, I’ll present TinyTL [3] that enables tiny transfer learning on-device, reducing the memory footprint by 7-13x.  Finally, I will describe data-efficient GAN training techniques [4] that can generate photo-realistic images using only 100 images, which used to require tens of thousands of images. We hope such TinyML techniques can make AI greener, faster, more efficient and more sustainable.

[1] MCUNet: Tiny Deep Learning on IoT Devices, (NeurIPS’20 spotlight)
[2] Once-for-All: Train One Network and Specialize it for Efficient Deployment (ICLR’19)
[3] Tiny Transfer Learning: Reduce Memory, not Parameters for Efficient On-Device Learning (NeurIPS’20)
[4] Differentiable Augmentation for Data-Efficient GAN Training (NeurIPS’20)

Bio:  Song Han is an assistant professor at MIT’s EECS. He received his PhD degree from Stanford University. His research focuses on efficient deep learning computing. He proposed “deep compression” technique that can reduce neural network size by an order of magnitude without losing accuracy, and the hardware implementation “efficient inference engine” that first exploited pruning and weight sparsity in deep learning accelerators. His team’s work on hardware-aware neural architecture search that bring deep learning to IoT devices was highlighted by MIT News, Wired, Qualcomm News, VentureBeat, IEEE Spectrum, integrated in PyTorch and AutoGluon, and received many low-power computer vision contest awards in flagship AI conferences (CVPR’19, ICCV’19 and NeurIPS’19). Song received Best Paper awards at ICLR’16 and FPGA’17, Amazon Machine Learning Research Award, SONY Faculty Award, Facebook Faculty Award, NVIDIA Academic Partnership Award. Song was named “35 Innovators Under 35” by MIT Technology Review for his contribution on “deep compression” technique that “lets powerful artificial intelligence (AI) programs run more efficiently on low-power mobile devices.” Song received the NSF CAREER Award for “efficient algorithms and hardware for accelerated machine learning” and the IEEE “AIs 10 to Watch: The Future of AI” award.


To request accommodations for a disability please contact Lori Bailey at lbailey@princeton.edu at least one week prior to the event.

Disruptive Research on Distributed ML Systems

Date and Time
Thursday, April 1, 2021 - 3:00pm to 4:00pm
Location
Zoom Webinar (off campus)
Type
Seminar
Host
Wyatt Lloyd, Princeton SNS group

Zoom link: https://princeton.zoom.us/j/906163778 


Guanhua Wang
Deep Neural Networks (DNNs) enable computers to excel across many different applications such as image classification, speech recognition and robotics control. To accelerate DNN training and serving, parallel computing is widely adopted. System efficiency is a big issue when scaling out. In this talk, I will make three arguments towards better system efficiency in distributed DNN training and serving.

First, Ring All-Reduce for model synchronization is not optimal, but Blink is. By packing spanning trees rather than forming rings, Blink achieves higher flexibility in arbitrary networking environments and provides near-optimal network throughput. Blink is filed as a US patent and is being used by Microsoft. Blink gains lots of attention from industry, such as Facebook (distributed PyTorch team), ByteDance (parent company of TikTok app). Blink was also featured on Nvidia GTC China 2019 and news from Baidu, Tencent, etc. 

Second, communication can be eliminated via sensAI's class parallelism. sensAI decouples a multi-task model into disconnected subnets, each is responsible for decision making of a single task. sensAI's attribute of low-latency, real-time model serving attracts several Venture Capitals in the Bay Area.

Third, Wavelet is more efficient than gang-scheduling. By intentionally adding task launching latency, Wavelet interleaves peak memory usage across different waves of training tasks on the accelerators, and thus it improves both computation and on-device memory utilization.

Bio: Guanhua Wang is a final year CS PhD in the RISELab at UC Berkeley, advised by Prof. Ion Stoica. His research lies primarily in the ML+System area including fast collective communication schemes for model synchronization, efficient in-parallel model training and real-time model serving.

Deep Learning: It’s Not All About Recognizing Cats and Dogs

Date and Time
Thursday, November 12, 2020 - 12:30pm to 1:30pm
Location
Zoom Webinar (off campus)
Type
Seminar
Host
Margaret Martonosi (CS) & David Wentzlaff (EE)

Carole-Jean Wu
Please register using this link.


In this seminar, we will examine 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, personalization and recommendation consumes even higher compute cycles of 80%. What does 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, IISWC-2020]. 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].
 
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 Computer Science and Electrical Engineering Korhammer Lecture Series Funds.

To request accommodations for a disability please contact Emily Lawrence, emilyl@cs.princeton.edu at least one week prior to the event.

High-speed network measurement under constrained programming model

Date and Time
Thursday, March 28, 2019 - 3:00pm to 4:00pm
Location
Zoom Meeting (off campus)
Type
Seminar

NPI Zoom Seminar-Xiaoqi (Danny) Chen

Network measurement is a vital tool for network operators to diagnose outages, optimize performance, and detect attacks. Recently, the development of programmable switches has enabled us to run measurement algorithms in the network switch directly, and analyze packets up to Tbps in throughput. However, the programming model of programmable switches is extremely constrained, which restricts the types of algorithms we can run in them.
 
To achieve much-needed measurement goals of network operators, we design tailored algorithms to adapt to the programming constraints imposed by practical programmable switches. We first present PRECISON, which attempts to answer the Heavy-Hitter Flow Detection problem (find out  which flows sent the largest number of packets). PRECISION can accurately identify the heavy-hitter
flows using a small amount of memory, by recirculating a small number of packets probabilistically. Then, we present Snappy, which tries to solve the Heavy Flow in the Queue problem (which flows occupied a large fraction of queuing buffer). Snappy can pinpoint the bursty flows causing ephemeral long queues and potential packet loss, by maintaining multiple traffic snapshots of short time intervals. Our measurement algorithms enable network operators to perform immediate actions against these specific network flows, inhibiting congestion in real-time, while potentially improving service quality for other network flows.
 
Bio:
Xiaoqi is a second year PhD student at Department of Computer Science, Princeton University,  advised by Prof. Jennifer Rexford. Before joining Princeton, he received his Bachelor's degree from  Institute for Interdisciplinary Information Sciences (Yao class), Tsinghua University. His research is  running network measurements in programmable switches. Interests also include data center  networking, sketches, and network science. 
----------------------------------------------------------------

Join Zoom Meeting
https://cornell.zoom.us/j/128236059

One tap mobile
+16468769923,,128236059# US (New York)
+16699006833,,128236059# US (San Jose)

Dial by your location
        +1 646 876 9923 US (New York)
        +1 669 900 6833 US (San Jose)
Meeting ID: 128 236 059
Find your local number: https://zoom.us/u/aB2V3pkFr 

Join by SIP
128236059@zoomcrc.com

Join by H.323
162.255.37.11 (US West)
162.255.36.11 (US East)
221.122.88.195 (China)
115.114.131.7 (India)
213.19.144.110 (EMEA)
202.177.207.158 (Australia)
209.9.211.110 (Hong Kong)
64.211.144.160 (Brazil)
69.174.57.160 (Canada)
Meeting ID: 128 236 059

Join by Skype for Business
https://cornell.zoom.us/skype/12823605

New Frontiers in Systems Theory: Cyber-Physical and Human Networks

Date and Time
Tuesday, September 18, 2018 - 4:30pm to 6:00pm
Location
Engineering Quadrangle B205
Type
Seminar
Speaker
Massimo Franceschetti, from University of California at San Diego

Abstract:
We illustrate through two examples some new challenges that network, information, and control theories are facing in a world of widespread sensing, ubiquitous connectivity, and availability of massive data. In the first part of the talk, we argue that in this new context the focus of network science should shift from studying the structural properties of the network to modeling the process dynamics occurring over the network, including the human dynamics. To elaborate on this point of view, we consider a Shelling-type model of agent dynamics leading to community formation in large networks that is related to the Ising model of statistical physics. We show how the model can be solved rigorously, after 50 years from its introduction, and what lessons can be learned from this new analysis. In the second part of the talk, we shift the focus from the analysis of dynamical systems to their control. We argue that in cyber-physical systems control must be performed over communication channels, and observations corrupted by noise are subject to delay and erasures. In this new framework, one of the central results is the data-rate theorem, describing the trade-off between the system dynamics and the data-rate available over the feedback loop. After reviewing this classic result, we show that the new paradigm of event-triggering control provides a completely new perspective, leading to new insights on the information-theoretic value of timing in communication.
 
Bio:
Massimo Franceschetti received the Laurea degree (with  honors) in computer engineering from the University of Naples, Naples, Italy, in 1997, the M.S. and Ph.D. degrees in ele trical engineering from the California Institute of Technology, Pasadena, CA, in 1999, and 2003, respectively. He is Professor of Electrical and Computer Engineering at the University of California at San Diego (UCSD). Before joining UCSD, he was a postdoctoral scholar at the University of California at Berkeley for two years.  He is co-author of the book “Random Networks for Communication” and author of the book “Wave theory of information,” both published by Cambridge University Press. Dr. Franceschetti served as Associate Editor for Communication Networks of the IEEE Transactions on Information Theory (2009 – 2012), as associate editor of the IEEE Transactions on Control of Network Systems (2013- 2016), as Associate Editor of the IEEE Transactions on Network Science and Engineering (2014-2017) and as Guest Associate Editor of the IEEE Journal on Selected Areas in Communications (2008, 2009).  He is an IEEE Fellow and was awarded the C. H. Wilts Prize in 2003 for best doctoral thesis in electrical engineering at Caltech; the S.A. Schelkunoff Award in 2005 for best paper in the IEEE Transactions on Antennas and Propagation, a National Science Foundation (NSF) CAREER award in 2006, an Office of Naval Research (ONR) Young Investigator Award in 2007, the IEEE Communications Society Best Tutorial Paper Award in 2010, and the IEEE Control theory society Ruberti young researcher award in 2012.

LIVING MATERIALS - Materials at the Interface Between Technology and Life Sciences

Date and Time
Tuesday, February 27, 2018 - 4:30pm to 5:30pm
Location
Friend Center 008
Type
Seminar
Speaker
Fiorenzo Omenetto, from Tufts University

EE Departmental Seminars

Abstract:
Structural proteins are Nature’s building blocks, conferring stiffness, structure, and function to ordinarily soft biological materials. Such proteins are polymorphic which allows controlling the end material format through their self-assembly. These biomaterials provide a unique opportunity by being simultaneously “technological” (e.g. mechanically robust, micro- and nanostructured, high-performing) and “biological” (e.g. living, adaptable, bio-functional) making them ideally suited for applications at the interface between these two domains. 
 
Bio:
Fiorenzo G. Omenetto is the Frank C. Doble Professor of Engineering, and a Professor of Biomedical Engineering at Tufts University.  He also holds appointments in the Department of Physics and the Department of Electrical Engineering. His research interests are at the interface of technology, biologically inspired materials and the natural sciences with an emphasis on new transformative approaches for sustainable materials for high-technology applications.  He also serves as Dean for Research for the School of Engineering. He has proposed and pioneered the use of silk as a material platform for advanced technology with uses in photonics, optoelectronics and nanotechnology applications, is co-inventor on several disclosures (~100) on the subject, and is actively investigating applications of this technology base both for technical and design applications.  He is a co-founder of three startups and has active roles in their governance.
Prof. Omenetto was formerly a J. Robert Oppenheimer Fellow at Los Alamos National Laboratories, a Guggenheim Fellow, and is a Fellow of the Optical Society of America and of the American Physical Society and a Senior Member of SPIE and is a Tallberg Foundation Global Leadership Fellow.  He was named one of the 50 top people in tech by Fortune magazine in a class including (among others) Steve Jobs, Jeff Bezos, Larry Page, Shigeru Miyamoto.  His research has been featured extensively in the press with coverage in the most important media outlets worldwide.

Harnessing Synthetic Quantum Matter

Date and Time
Monday, February 26, 2018 - 4:30pm to 5:30pm
Location
Engineering Quadrangle B205
Type
Seminar
Speaker
Alexey Gorshkov, from University of Maryland

EE Departmental Seminars

Abstract:
Recent advances in condensed matter, optical, and atomic physics led to the emergence of highly controllable synthetic quantum matter, such as superconducting circuits, implanted solid-state defects, trapped atoms or ions, and strongly interacting photons. In addition to allowing us to gain fundamental insights into peculiar and diverse behavior of many-body -- that is, large and interacting -- quantum systems, synthetic quantum matter paves the way for building revolutionary quantum technologies such as extraordinarily powerful computers, unbreakably secure communication devices, and exceptionally accurate sensors. In this talk, we will explore two facets of synthetic quantum matter. First, we will argue that sampling complexity, that is the question of how hard it is to produce a sample from a given probability distribution, lies at the heart of understanding and harnessing synthetic quantum matter. Second, we will show how to engineer interactions between individual photons and use these interactions for building quantum technologies and accessing exotic few-body and many-body physics. Finally, we will put this work in the context of a broader quest to design, understand, control, and harness synthetic quantum matter. 

Bio:
Alexey Gorshkov received his A.B. and Ph.D. degrees from Harvard in 2004 and 2010, respectively. In 2013, after three years as a Lee A. DuBridge Postdoctoral Scholar at Caltech, he became a staff physicist at NIST. At the same time, he started his own research group at the University of Maryland, where he is a fellow of the Joint Quantum Institute and of the Joint Center for Quantum Information and Computer Science. His theoretical research is at the interface of quantum optics, atomic physics, condensed matter physics, and quantum information science. Applications of his research include quantum computing, quantum communication, and quantum sensing.

You have to be brilliant to do that! Cultures of Brilliance and Academic Gender Gaps

Date and Time
Wednesday, February 3, 2016 - 4:00pm to 5:00pm
Location
Andlinger Center Maeder Hall
Type
Seminar
Host
SEAS-sponsored Seminar, http://www.princeton.edu/engineering/

You have to be brilliant to do that!
Many academic fields, including engineering and computer science, have persistent gender gaps despite attempts to encourage women’s participation. While this is often characterized as a problem faced by STEM fields, some humanities disciplines such as philosophy face precisely the same challenges, while some STEM disciplines such as molecular biology are far more gender-balanced. Are there any isolable factors that predict the occurrence of gender gaps across the entire academic spectrum? This talk presents data suggesting that one such factor may be academics' beliefs about what is required for success in their discipline. In some fields, success is viewed primarily as a matter of hard work and dedication -- but in others, success is seen as requiring a special, unteachable spark of brilliance. Cultural stereotypes strongly associate this sort of raw genius with men rather than women -- where are the female Sherlock Holmes, Dr. Houses, or Will Huntings? -- meaning that women may be discouraged from pursuing careers in such fields. These findings point to new strategies for increasing women’s participation in STEM, and indicate concrete measures that might be implemented in SEAS classrooms to better include and encourage women students.

Interference Rendered Significantly Harmless

Date and Time
Friday, December 12, 2008 - 1:00pm to 2:30pm
Location
Peyton Hall 145
Type
Seminar
Speaker
Ramakrishna Gummadi, from MIT
Host
Jennifer Rexford
Abstract: The throughput of existing wireless networks is often limited by interference. One fundamental reason is that the current designs are constrained by a "one-transmission-at-a-time" model at the link layer and a fixed-width spectrum allocation at the physical layer. We present a new wireless design that exploits traffic burstiness and node heterogeneity, thereby improving concurrency and spectrum usage. The main challenge is the unmanaged nature of many wireless networks such as 802.11 and mesh, which makes centralized resource allocation impractical. We show through analysis and implementation that simple randomized allocation policies can overcome this challenge, and improve throughput by 2x or more.

This work is joint with Rabin Patra, Hari Balakrishnan and Eric Brewer.

Bio: Ramakrishna Gummadi is a post-doc at the MIT Computer Science and Artificial Intelligence Laboratoray (CSAIL). He obtained his B.Tech. from IIT-Madras, M.S. from UC Berkeley and Ph.D. from USC, all in Computer Science. His dissertation was about reliable and efficient programming languages for sensor networks. He is interested in building scalable and reliable systems and networks based on sound principles. His awards include a UC Regents Graduate Fellowship, a best paper awarded out of all 2001 Journal of Computer Networks papers, a best poster/demo award at SenSys 2004, and an award at the ACM Student Research Competition (SRC) held at PLDI 2007.

Follow us: Facebook Twitter Linkedin