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

Fwd: Colloquium speaker Silvio Savarese Today, 4:30pm
by Adam Finkelstein 09 Dec '14

09 Dec '14

Silvio Savarese talk starting momentarily in the small auditorium in the CS Building... ---------- Forwarded message ---------- From: Nicole E. Wagenblast <nwagenbl(a)cs.princeton.edu> Date: Tue, Dec 9, 2014 at 9:01 AM Subject: [talks] Colloquium speaker Silvio Savarese Today, 4:30pm To: "Talks (colloquium)" <talks(a)lists.cs.princeton.edu> Colloquium Speaker Silvio Savarese, Stanford University Tuesday, December 9, 4:30pm Computer Science 105 Sensing and sensibility We are moving into a world where sensors will be everywhere and will be ubiquitous. In the recent years, we have seen an explosion of new artificial visual sensors that can integrate luminance values with other sensing modalities: infrared, thermal, gravity, depth, to cite a few. Sensing is not the hard problem here, however. Sensibility, or, intelligent understanding of the sensing data is the challenge. When we look at an environment such as a coffee shop, we don't just recognize the objects in isolation, but rather perceive a rich scenery of the 3D space, its objects and all the relations among them. This allows us to effortlessly navigate through the environment, or to interact and manipulate objects in the scene with amazing precision. A major line of work from my group in recent years has been to design visual models that can process different sensing modalities and enable sensibility. I will start by giving an overview of our research for detecting objects and determining their geometric and physical properties such as 3D location, pose or shape from images, videos or RGB-Depth imagery. Then, I will demonstrate that these detection methods play a critical role for modeling the interplay between objects and space which, in turn, enable simultaneous inference of the semantic and spatial properties of a scene. I will conclude this talk by demonstrating that our models for sense and sensibility are potentially transformative in application areas related to autonomous or assisted navigation, smart environments, augmented reality, and large scale information management. Silvio Savarese is an Assistant Professor of Computer Science at Stanford University. He earned his Ph.D. in Electrical Engineering from the California Institute of Technology in 2005 and was a Beckman Institute Fellow at the University of Illinois at Urbana-Champaign from 2005-2008. He joined Stanford in 2013 after being Assistant and then Associate Professor (with tenure) of Electrical and Computer Engineering at the University of Michigan, Ann Arbor, from 2008 to 2013. His research interests include computer vision, object recognition and scene understanding, shape representation and reconstruction, human activity recognition and visual psychophysics. He is recipient of several awards including the James R. Croes Medal in 2013, a TRW Automotive Endowed Research Award in 2012, an NSF Career Award in 2011 and Google Research Award in 2010. In 2002 he was awarded the Walker von Brimer Award for outstanding research initiative.

1 0

Prem Gopalan will present his FPO on December 15, 2014 at 2pm in CS 402
by Nicki Gotsis 09 Dec '14

09 Dec '14

Prem Gopalan will present his Final Public Oral Examination December 15, 2014 at 2pm in 402 Computer Science Building. The members of his committee are: Michael Joseph Freedman, non reader; John D. Storey, non reader (Molecular Biology, Princeton); and Jake Hofman, Reader (Microsoft Research). A copy of his thesis is available in Room 310. Everyone is invited to attend his talk. Nicki Gotsis Graduate Coordinator Room 310, Computer Science (609) 258-5387 ngotsis(a)cs.princeton.edu

1 1

Colloquium speaker Silvio Savarese Tuesday Dec 9, 4:30pm
by Nicole E. Wagenblast 09 Dec '14

09 Dec '14

Colloquium Speaker Silvio Savarese, Stanford University Tuesday, December 9, 4:30pm Computer Science 105 Sensing and sensibility We are moving into a world where sensors will be everywhere and will be ubiquitous. In the recent years, we have seen an explosion of new artificial visual sensors that can integrate luminance values with other sensing modalities: infrared, thermal, gravity, depth, to cite a few. Sensing is not the hard problem here, however. Sensibility, or, intelligent understanding of the sensing data is the challenge. When we look at an environment such as a coffee shop, we don't just recognize the objects in isolation, but rather perceive a rich scenery of the 3D space, its objects and all the relations among them. This allows us to effortlessly navigate through the environment, or to interact and manipulate objects in the scene with amazing precision. A major line of work from my group in recent years has been to design visual models that can process different sensing modalities and enable sensibility. I will start by giving an overview of our research for detecting objects and determining their geometric and physical properties such as 3D location, pose or shape from images, videos or RGB-Depth imagery. Then, I will demonstrate that these detection methods play a critical role for modeling the interplay between objects and space which, in turn, enable simultaneous inference of the semantic and spatial properties of a scene. I will conclude this talk by demonstrating that our models for sense and sensibility are potentially transformative in application areas related to autonomous or assisted navigation, smart environments, augmented reality, and large scale information management. Silvio Savarese is an Assistant Professor of Computer Science at Stanford University. He earned his Ph.D. in Electrical Engineering from the California Institute of Technology in 2005 and was a Beckman Institute Fellow at the University of Illinois at Urbana-Champaign from 2005–2008. He joined Stanford in 2013 after being Assistant and then Associate Professor (with tenure) of Electrical and Computer Engineering at the University of Michigan, Ann Arbor, from 2008 to 2013. His research interests include computer vision, object recognition and scene understanding, shape representation and reconstruction, human activity recognition and visual psychophysics. He is recipient of several awards including the James R. Croes Medal in 2013, a TRW Automotive Endowed Research Award in 2012, an NSF Career Award in 2011 and Google Research Award in 2010. In 2002 he was awarded the Walker von Brimer Award for outstanding research initiative.

1 1

Dan Larkin will present his Pre FPO on Tuesday, December 9 at 11am in CS 302.
by Nicki Gotsis 08 Dec '14

08 Dec '14

Dan Larkin will present his Pre FPO on Tuesday, December 9, 2014 at 11am in CS 302. The members of his committee are: Robert Tarjan (advisor), Jeff Erickson and Robert Sedgewick (readers), and Mark Braverman and Sanjeev Arora. Everyone is invited to attend his talk. The talk title and abstract follow below: Title: Compressing Trees with a Sledgehammer Abstract: A number of fundamental algorithms make use of the compressed tree framework to accelerate computation on graphs and sets. Perhaps the most well-known of these problems is disjoint set union (union-find). The main contributions of this disseration are a simple, optimal, randomized solution to the disjoint set union problem; an optimal, space-saving solution to the nested variation of the same problem; and a clean, unified analysis which encapsulates virtually all known optimal compressed tree algorithms and provides significant flexibility to handle a wide class of possible algorithms. Nicki Gotsis Graduate Coordinator Room 310, Computer Science (609) 258-5387 ngotsis(a)cs.princeton.edu

1 1

Colloquium Speaker Thomas Wenisch, Today, Dec 8, 4:30pm
by Nicole E. Wagenblast 08 Dec '14

08 Dec '14

Colloquium Speaker Thomas Wenisch, University of Michigan Monday, December 8, 4:30pm Computer Science 105 Computational Sprinting Although transistor density continues to increase, power density is increasing each chip generation. Particularly in mobile devices, which have limited cooling options, these trends lead to "dark silicon" in which sustained chip performance is limited primarily by power rather than area. However, many mobile applications do not demand sustained performance; rather they comprise short bursts of computation in response to sporadic user activity. To improve responsiveness for such applications, this talk describes "computational sprinting ", in which the system activates otherwise powered-down cores for brief bursts of intense parallel computation. During a sprint the chip temporarily exceeds its sustainable thermal power budget to provide instantaneous throughput, after which the chip must return to nominal operation to cool down. To enable longer sprints , we describe a thermal design that incorporates phase-change materials to provide thermal capacitance. Results indicate that sprinting can both provide large improvements in responsiveness while actually saving energy due to race-to-idle effects. Thomas Wenisch is an Associate Professor of Computer Science and Engineering at the University of Michigan, specializing in computer architecture. His prior research includes memory streaming for commercial server applications, multiprocessor memory systems, memory disaggregation, and rigorous sampling-based performance evaluation methodologies. His ongoing work focuses on computational sprinting , server and data center architectures, programming models for byte-addressable NVRAM, and architectures to enable hand-held 3D ultrasound . Wenisch received the NSF CAREER award in 2009. Prior to his academic career, Wenisch was a software developer at American Power Conversion, where he worked on data center thermal topology estimation. He received his Ph.D. in Electrical and Computer Engineering from Carnegie Mellon University.

1 0

Colloquium Speaker Thomas Wenisch, Monday, Dec 8, 4:30pm
by Nicole E. Wagenblast 07 Dec '14

07 Dec '14

Colloquium Speaker Thomas Wenisch, University of Michigan Monday, December 8, 4:30pm Computer Science 105 Computational Sprinting Although transistor density continues to increase, power density is increasing each chip generation. Particularly in mobile devices, which have limited cooling options, these trends lead to "dark silicon" in which sustained chip performance is limited primarily by power rather than area. However, many mobile applications do not demand sustained performance; rather they comprise short bursts of computation in response to sporadic user activity. To improve responsiveness for such applications, this talk describes "computational sprinting ", in which the system activates otherwise powered-down cores for brief bursts of intense parallel computation. During a sprint the chip temporarily exceeds its sustainable thermal power budget to provide instantaneous throughput, after which the chip must return to nominal operation to cool down. To enable longer sprints , we describe a thermal design that incorporates phase-change materials to provide thermal capacitance. Results indicate that sprinting can both provide large improvements in responsiveness while actually saving energy due to race-to-idle effects. Thomas Wenisch is an Associate Professor of Computer Science and Engineering at the University of Michigan, specializing in computer architecture. His prior research includes memory streaming for commercial server applications, multiprocessor memory systems, memory disaggregation, and rigorous sampling-based performance evaluation methodologies. His ongoing work focuses on computational sprinting , server and data center architectures, programming models for byte-addressable NVRAM, and architectures to enable hand-held 3D ultrasound . Wenisch received the NSF CAREER award in 2009. Prior to his academic career, Wenisch was a software developer at American Power Conversion, where he worked on data center thermal topology estimation. He received his Ph.D. in Electrical and Computer Engineering from Carnegie Mellon University.

1 0

Colloquium speaker Kristen Grauman, Today, 4:30pm
by Nicole E. Wagenblast 03 Dec '14

03 Dec '14

Colloquium Speaker Kristen Grauman, University of Texas at Austin Wednesday, December 3, 4:30pm Computer Science 105 Searching and Browsing Visual Data Widespread visual sensors and unprecedented connectivity have left us awash with visual data---from online photo collections, home videos, news footage, medical images, or surveillance feeds. How can we efficiently browse image and video collections based on semantically meaningful criteria? How can we bring order to the data, beyond manually defined keyword tags? I will present work exploring these questions in the context of interactive visual search and summarization. In particular, I’ll first introduce attribute representations that connect visual properties to human describable terms. I’ll show how these attributes enable both fine-grained content-based retrieval as well as new forms of human supervision for recognition problems. Then, I’ll overview our recent work on video summarization, where the goal is to automatically transform a long video into a short one. Using videos captured with egocentric wearable cameras, we’ll see how hours of data can be distilled to a succinct visual storyboard that is understandable in just moments. Together, these ideas are promising steps towards widening the channel of communication between humans and computer vision algorithms, which is critical to facilitate efficient browsing of large-scale image and video collections. This is work done with Adriana Kovashka, Yong Jae Lee, Devi Parikh, Lu Zheng, Bo Xiong, and Dinesh Jayaraman. Kristen Grauman is an Associate Professor in the Department of Computer Science at the University of Texas at Austin. Her research in computer vision and machine learning focuses on visual search and object recognition. Before joining UT-Austin in 2007, she received her Ph.D. in the EECS department at MIT, in the Computer Science and Artificial Intelligence Laboratory. She is an Alfred P. Sloan Research Fellow and Microsoft Research New Faculty Fellow, a recipient of NSF CAREER and ONR Young Investigator awards, the Regents' Outstanding Teaching Award from the University of Texas System in 2012, the PAMI Young Researcher Award in 2013, the 2013 Computers and Thought Award from the International Joint Conference on Artificial Intelligence, and a Presidential Early Career Award for Scientists and Engineers (PECASE) in 2013. She and her collaborators were recognized with the CVPR Best Student Paper Award in 2008 for their work on hashing algorithms for large-scale image retrieval, and the Marr Best Paper Prize at ICCV in 2011 for their work on modeling relative visual attributes.

1 0

Colloquium Speaker Thomas Wenisch, Monday, Dec 8, 4:30pm
by Nicole E. Wagenblast 02 Dec '14

02 Dec '14

Colloquium Speaker Thomas Wenisch, University of Michigan Monday, December 8, 4:30pm Computer Science 105 Computational Sprinting Although transistor density continues to increase, power density is increasing each chip generation. Particularly in mobile devices, which have limited cooling options, these trends lead to "dark silicon" in which sustained chip performance is limited primarily by power rather than area. However, many mobile applications do not demand sustained performance; rather they comprise short bursts of computation in response to sporadic user activity. To improve responsiveness for such applications, this talk describes "computational sprinting ", in which the system activates otherwise powered-down cores for brief bursts of intense parallel computation. During a sprint the chip temporarily exceeds its sustainable thermal power budget to provide instantaneous throughput, after which the chip must return to nominal operation to cool down. To enable longer sprints , we describe a thermal design that incorporates phase-change materials to provide thermal capacitance. Results indicate that sprinting can both provide large improvements in responsiveness while actually saving energy due to race-to-idle effects. Thomas Wenisch is an Associate Professor of Computer Science and Engineering at the University of Michigan, specializing in computer architecture. His prior research includes memory streaming for commercial server applications, multiprocessor memory systems, memory disaggregation, and rigorous sampling-based performance evaluation methodologies. His ongoing work focuses on computational sprinting , server and data center architectures, programming models for byte-addressable NVRAM, and architectures to enable hand-held 3D ultrasound . Wenisch received the NSF CAREER award in 2009. Prior to his academic career, Wenisch was a software developer at American Power Conversion, where he worked on data center thermal topology estimation. He received his Ph.D. in Electrical and Computer Engineering from Carnegie Mellon University.

1 0

Colloquium speaker Kristen Grauman, Wed Dec 3, 4:30pm
by Nicole E. Wagenblast 02 Dec '14

02 Dec '14

Colloquium Speaker Kristen Grauman, University of Texas at Austin Wednesday, December 3, 4:30pm Computer Science 105 Searching and Browsing Visual Data Widespread visual sensors and unprecedented connectivity have left us awash with visual data---from online photo collections, home videos, news footage, medical images, or surveillance feeds. How can we efficiently browse image and video collections based on semantically meaningful criteria? How can we bring order to the data, beyond manually defined keyword tags? I will present work exploring these questions in the context of interactive visual search and summarization. In particular, I’ll first introduce attribute representations that connect visual properties to human describable terms. I’ll show how these attributes enable both fine-grained content-based retrieval as well as new forms of human supervision for recognition problems. Then, I’ll overview our recent work on video summarization, where the goal is to automatically transform a long video into a short one. Using videos captured with egocentric wearable cameras, we’ll see how hours of data can be distilled to a succinct visual storyboard that is understandable in just moments. Together, these ideas are promising steps towards widening the channel of communication between humans and computer vision algorithms, which is critical to facilitate efficient browsing of large-scale image and video collections. This is work done with Adriana Kovashka, Yong Jae Lee, Devi Parikh, Lu Zheng, Bo Xiong, and Dinesh Jayaraman. Kristen Grauman is an Associate Professor in the Department of Computer Science at the University of Texas at Austin. Her research in computer vision and machine learning focuses on visual search and object recognition. Before joining UT-Austin in 2007, she received her Ph.D. in the EECS department at MIT, in the Computer Science and Artificial Intelligence Laboratory. She is an Alfred P. Sloan Research Fellow and Microsoft Research New Faculty Fellow, a recipient of NSF CAREER and ONR Young Investigator awards, the Regents' Outstanding Teaching Award from the University of Texas System in 2012, the PAMI Young Researcher Award in 2013, the 2013 Computers and Thought Award from the International Joint Conference on Artificial Intelligence, and a Presidential Early Career Award for Scientists and Engineers (PECASE) in 2013. She and her collaborators were recognized with the CVPR Best Student Paper Award in 2008 for their work on hashing algorithms for large-scale image retrieval, and the Marr Best Paper Prize at ICCV in 2011 for their work on modeling relative visual attributes.

1 0

Cole Schlesinger will present his Pre-FPO Tuesday, December 9 in CS 402 at 10am (Title: Abstractions for Software-defined Networks)
by Nicki Gotsis 02 Dec '14

02 Dec '14

Cole Schlesinger will present his Pre-FPO Tuesday, December 9, 2014 in CS 402 at 10am. The member of his committee are: Dave Walker (advisor), Aarti Gupta and Arjun Guha (readers), Nick Feamster and Jen Rexford (non-readers). Everyone is invited to attend his talk. The talk title and abstract follow below: Title: Abstractions for Software-defined Networks Abstract: In a Software-Defined Network (SDN), a central, computationally powerful controller manages a set of distributed, computationally simple switches. The controller computes a policy describing how each switch should route packets and populates packet-processing tables on each switch with rules to enact the routing policy. As network conditions change, the controller continues to add and remove rules from switches to adjust the policy as needed. My research focuses on defining new, high-level languages and abstractions for expressing network policy; defining technical mechanisms for reasoning formally about those abstractions; and defining algorithms for compiling high-level abstractions into available low-level mechanisms. In the first part of this talk, I present NetKAT, a language for expressing network-wide packet-forwarding policies. NetKAT comes equipped with a denotational semantics and a sound and complete equational theory drawn from a well-studied mathematical structure called a Kleene algebra with tests (KAT). The equational theory has many practical applications, such as syntactic techniques for checking reachability and proving non-interference properties that ensure isolation between policies. Here, I focus on applying the equational theory to guide and establish the correctness of a compilation algorithm targeting switches that expose simple, single-table packet-processing pipelines. In the second part of this talk, I show how to extend the switch-specific fragment of NetKAT with operators for writing and deploying network policies to switches with multiple-table, reconfigurable packet-processing pipelines. The language, called Concurrent NetCore, allows a network programmer to specify a virtual packet-processing pipeline, an abstraction that captures the physical resources required by the controller and also hides the details of the (potentially varied) pipelines of switches in the network. I also show how to use Concurrent NetCore to model the pipelines of three switching architectures and present compilation algorithms for mapping arbitrary virtual pipelines---and arbitrary packet-processing policies---to the physical pipelines. Along the way, I explore metatheoretical properties of the language and present correctness proofs for the compilation algorithms. Together, these abstractions and compilation algorithms pave the way for easier fine-grained network management. This talk draws on joint work with Carolyn Jane Anderson, Nate Foster, Michael Greenberg, Arjun Guha, Jean-Baptiste Jeannin, Dexter Kozen, and David Walker. Thanks! Cole

1 0