- talks - lists.cs.princeton.edu

Joshua Kroll will present his FPO, "Accountable Algorithms" on Friday, August 21, 2015 at 11am in Sherrerd Hall 306.
by Nicki Gotsis 12 Aug '15

12 Aug '15

Joshua Kroll will present his FPO, "Accountable Algorithms" on Friday, August 21, 2015 at 11am in Sherrerd Hall 306. The members of his committee are: Readers: Arvind Narayanan, Matthew Green (Johns Hopkins University) Examiners: Edward Felten (adviser), Andrew Appel, and Nicholas Feamster A copy of his thesis is available in Room 310. Everyone is invited to attend his talk.

1 0

EE Seminar - August 10, 2015, 11:00 am, E-Quad B205 - Prof. Gernot Heiser
by Nicole E. Wagenblast 30 Jul '15

30 Jul '15

SeminarLogo.jpg Speaker: Prof. Gernot Heiser University of New South Wales Title: Towards Principled Temporal Isolation in the Operating System Date: Monday, August 10, 2015 Time: 11:00 a.m. Room: E-Quad, Room B205 Host: Prof. Ruby Lee Abstract: Spatial isolation, i.e. controlling memory access, forms the basis of protection in operating systems. However, this is neither sufficient for ensuring safe operation of real-time systems, nor for ensuring confidentiality. It must be complemented by temporal isolation, i.e. preventing a process from unduly influencing the execution speed of another process. Most operating systems provide very poor temporal isolation if any. Exceptions are systems designed to support a narrow class of applications (either hard real-time systems or separation kernels for security-critical uses), which also tend to have high overheads or force low system utilisation. The formally-verified seL4 microkernel is a general-purpose high-assurance platform that is intended to support a wide range of uses, including highly performance-sensitive mobile systems, (safety) multi-criticality systems and (security-critical) cross-domain solutions. We are therefore after mechanisms that provide strong temporal isolation where needed, but avoid overheads where not. The talk will discuss present work in progress on simple kernel mechanisms that will allow us to achieve those goals. Biography: Gernot Heiser is Scientia Professor and John Lions Chair of Operating Systems at UNSW, and leads the Software Systems Research Group at NICTA, Australia’s National Centre of Excellence for ICT Research. He joined NICTA at its creation in 2002, and before that was a full-time member of academic staff at UNSW from 1991. His past work included the Mungi single-address-space operating system (OS), several un-broken records in IPC performance, and the best-ever reported performance for user-level device drivers. More recently he led the team at NICTA which produced the world’s first formal proof of functional correctness of a protected general-purpose operating-system kernel, the first published sound worst-case execution time analysis of a protected OS kernel, and the synthesis of high-performance device drivers. In 2006, Gernot with a number of his students founded Open Kernel Labs, to market secure virtualization technology for embedded systems. The company’s OKL4 operating system, a descendent of L4 kernels developed by his group at UNSW and NICTA, is deployed in billions of mobile devices, and now ships on the security processor of all iOS devices. Open Kernel Labs was sold to General Dynamics in August 2012.

1 0

11am Mon Aug 10 talk in E-Quad B205 on principled temporal isolation in operating systems
by Jennifer Rexford 30 Jul '15

30 Jul '15

> From: "Lori A. Bailey" <lbailey(a)Princeton.EDU> > Subject: EE Seminar - August 10, 2015, 11:00 am, E-Quad B205 - Prof. Gernot Heiser > Date: July 30, 2015 at 11:42:49 AM EDT > To: ee-seminar(a)Princeton.EDU > Reply-To: "Lori A. Bailey" <lbailey(a)Princeton.EDU> > > > Speaker: Prof. Gernot Heiser > University of New South Wales > Title: Towards Principled Temporal Isolation in the Operating System > Date: Monday, August 10, 2015 > Time: 11:00 a.m. > Room: E-Quad, Room B205 > Host: Prof. Ruby Lee > > Abstract: Spatial isolation, i.e. controlling memory access, forms the basis of protection in operating systems. However, this is neither sufficient for ensuring safe operation of real-time systems, nor for ensuring confidentiality. It must be complemented by temporal isolation, i.e. preventing a process from unduly influencing the execution speed of another process. Most operating systems provide very poor temporal isolation if any. Exceptions are systems designed to support a narrow class of applications (either hard real-time systems or separation kernels for security-critical uses), which also tend to have high overheads or force low system utilisation. > > The formally-verified seL4 microkernel is a general-purpose high-assurance platform that is intended to support a wide range of uses, including highly performance-sensitive mobile systems, (safety) multi-criticality systems and (security-critical) cross-domain solutions. We are therefore after mechanisms that provide strong temporal isolation where needed, but avoid overheads where not. The talk will discuss present work in progress on simple kernel mechanisms that will allow us to achieve those goals. > > > Biography: Gernot Heiser is Scientia Professor and John Lions Chair of Operating Systems at UNSW, and leads the Software Systems Research Group at NICTA, Australia’s National Centre of Excellence for ICT Research. He joined NICTA at its creation in 2002, and before that was a full-time member of academic staff at UNSW from 1991. His past work included the Mungi single-address-space operating system (OS), several un-broken records in IPC performance, and the best-ever reported performance for user-level device drivers. More recently he led the team at NICTA which produced the world’s first formal proof of functional correctness of a protected general-purpose operating-system kernel, the first published sound worst-case execution time analysis of a protected OS kernel, and the synthesis of high-performance device drivers. > > In 2006, Gernot with a number of his students founded Open Kernel Labs, to market secure virtualization technology for embedded systems. The company’s OKL4 operating system, a descendent of L4 kernels developed by his group at UNSW and NICTA, is deployed in billions of mobile devices, and now ships on the security processor of all iOS devices. Open Kernel Labs was sold to General Dynamics in August 2012. > > > > > > . > >

1 0

Haipeng Luo will present his Pre FPO on July 27, 2015 at 2pm in CS 401
by Nicki Gotsis 27 Jul '15

27 Jul '15

Haipeng Luo will present his Pre FPO on July 27, 2015 at 2pm in CS 401. The committee members are Robert Schapire (adviser); Elad Hazan and Readers: Satyen Kale (Yahoo Labs); and Non-readers: Sanjeev Arora, Barbara Engelhardt Everyone is invited to attend his talk. The talk title and abstract follow below: Title: Optimal and Adaptive Online Learning Online learning is one of the important and well-established learning models in the area of machine learning. Generally speaking, the goal of online learning is to make a sequence of accurate predictions â€œon the flyâ€ when interacting with the environment. Online learning has been extensively studied in recent years, and has also become of great interest to practitioners due to its applicability to large scale applications such as advertisement placement and recommendation systems. Existing online learning algorithms, however, are not always directly applicable in practice, because they usually rely on sophisticated tuning of parameters that gives sound theoretical guarantees but hardly works in practice. In this thesis, our main focus is thus to design practical and ready-to-use online learning algorithms that are parameter-free and work simultaneously under different patterns of data as well as different evaluation criteria. Our main contributions are new optimal and adaptive algorithms for the following three online learning problems: 1. We study the classic problem of online learning by combining expert advice with several new and challenging goals. By converting the problem into a â€œdrifting gameâ€ and analyzing the approximate minimax optimal solutions, we propose a new parameter-free algorithm called AdaNormalHedge. Our algorithm is the first one that achieves several desirable objectives simultaneously, including learning from unknown or even infinite number of experts, competing with a sequence of unknown competitors, and achieving much better performance when dealing with easy problems, while still ensuring worst case robustness in difficult situations. Our results lead to significant improvements in several applications, such as predicting almost as well as the best pruning tree of a decision tree. 2. The second problem we study is online boosting. Boosting is one of the central methodology in machine learning that studies the task of combining weak rule-of-thumbs into a strong learning algorithm. While it is well-studied in the offline batch setting, there is very few work on the theory of online boosting. We thus develop a set of optimal and adaptive online boosting algorithms, some that achieve arbitrarily small error under a novel notion of weak online learnability, and the others which converge to the best possible achieved by the linear combination of weak learners. Experiments on benchmark data are conducted to show that our new algorithms indeed improve upon previous work. 3. The last problem we investigate is online learning without knowing the actual time horizon, which is yet another important feature for a practical algorithm. Previously, different techniques have been used, such as the so-called â€œdoubling trickâ€ which is unfortunately inelegant and wasteful. We formalize this problem as a game and study the exact minimax optimal solutions. We first prove that learning with unknown time horizon is strictly harder, and then propose a simple minimax optimal algorithm for a special case. Based on this special case, we further develop a new algorithm for the general case which, although not being exact optimal, enjoys a better performance guarantee compared to previous work.

2 2

Taewook Oh will present his FPO on Tuesday, 7/28/2015 at 9:30am, 401 Computer Science
by Nicki Gotsis 20 Jul '15

20 Jul '15

Taewook Oh will present his FPO, "AUTOMATIC EXPLOITATION OF INPUT PARALLELISM, on Tuesday, 7/28/2015 at 9:30am, 401 Computer Science The members of his committee are Prof. David August (advisor); Readers: Prof. Brian Kernighan, Prof. Scott Mahlke (Univ. of Michigan) ; Non-readers: Prof. Andrew Appel, Prof. Sharad Malik (ELE). A copy of his thesis is available in Room 310. Everyone is invited to attend his talk.

1 0

today: 11am talk on measuring Internet quality of experience
by Jennifer Rexford 20 Jul '15

20 Jul '15

Speaker: Renata Teixeira, Inria Paris-Rocquencourt Title: HostView: Measuring Internet quality of experience on end-hosts Date/time: 11:00am-12:00pm Monday July 20 Location: CS 402 Abstract: There is much interest recently in doing automated performance diagnosis on user laptops or desktops. One interesting aspect of performance diagnosis that has received little attention is the user perspective on performance. To conduct research on both end-host performance diagnosis and user perception of network and application performance, we designed an end-host data collection tool, called HostView. HostView not only collects network, application and machine level data, but also gathers feedback directly from users. User feedback is obtained via two mechanisms, a system-triggered questionnaire and a user-triggered feedback form, that for example asks users to rate the performance of their network and applications. This talk will describe our experience with the first deployment of HostView. Using data from 40 users, we articulate the challenges in this line of research, and report on initial findings in correlating user data to system-level data. We then describe our more recent efforts in conducting an in-depth study with twelve users in France to guide our design of the next version of HostView and of methods to infer user context and activities. This is joint work with Diana Joumblatt, Jaideep Chandrashekar, Nina Taft, Anna-Kaisa Pietilainen, George Rosca, Peter Tolmie, Tom Rodden, Tom Lodge, Richard Motier. This work is funded in part by the EU FP7 User-Centric Networking project, Grant No. 611001. Bio: Renata Teixeira received the Ph.D. degree in computer science from the University of California, San Diego, in 2005. During her Ph.D. studies, she worked on Internet routing at the AT&T Research. She was a researcher with the Centre National de la Recherche Scientifique (CNRS) at LIP6, UPMC Sorbonne Universites, Paris, France from 2006 to 2013. She joined Inria Paris-Rocquencourt as senior researcher in October 2013. She was a visiting scholar at UC Berkeley/ICSI in 2011. Her research interests are in measurement, analysis, and management of data networks. She has authored more than 60 papers in this area. Renata is vice-chair of ACM SIGCOMM. She is a member of the steering committee of the ACM Internet Measurement Conference and has been active in the program committees of ACM SIGCOMM, ACM IMC, ACM CoNEXT, PAM, IEEE INFOCOM, among others.

1 0

11am Mon Jul 20 talk on measuring Internet quality of experience
by Jennifer Rexford 16 Jul '15

16 Jul '15

[Please contact Jen Rexford at jrex(a)cs.princeton.edu <mailto:jrex@cs.princeton.edu> if you are interested in meeting with the speaker.] Speaker: Renata Teixeira, Inria Paris-Rocquencourt Title: HostView: Measuring Internet quality of experience on end-hosts Date/time: 11:00am-12:00pm Monday July 20 Location: CS 402 Abstract: There is much interest recently in doing automated performance diagnosis on user laptops or desktops. One interesting aspect of performance diagnosis that has received little attention is the user perspective on performance. To conduct research on both end-host performance diagnosis and user perception of network and application performance, we designed an end-host data collection tool, called HostView. HostView not only collects network, application and machine level data, but also gathers feedback directly from users. User feedback is obtained via two mechanisms, a system-triggered questionnaire and a user-triggered feedback form, that for example asks users to rate the performance of their network and applications. This talk will describe our experience with the first deployment of HostView. Using data from 40 users, we articulate the challenges in this line of research, and report on initial findings in correlating user data to system-level data. We then describe our more recent efforts in conducting an in-depth study with twelve users in France to guide our design of the next version of HostView and of methods to infer user context and activities. This is joint work with Diana Joumblatt, Jaideep Chandrashekar, Nina Taft, Anna-Kaisa Pietilainen, George Rosca, Peter Tolmie, Tom Rodden, Tom Lodge, Richard Motier. This work is funded in part by the EU FP7 User-Centric Networking project, Grant No. 611001. Bio: Renata Teixeira received the Ph.D. degree in computer science from the University of California, San Diego, in 2005. During her Ph.D. studies, she worked on Internet routing at the AT&T Research. She was a researcher with the Centre National de la Recherche Scientifique (CNRS) at LIP6, UPMC Sorbonne Universites, Paris, France from 2006 to 2013. She joined Inria Paris-Rocquencourt as senior researcher in October 2013. She was a visiting scholar at UC Berkeley/ICSI in 2011. Her research interests are in measurement, analysis, and management of data networks. She has authored more than 60 papers in this area. Renata is vice-chair of ACM SIGCOMM. She is a member of the steering committee of the ACM Internet Measurement Conference and has been active in the program committees of ACM SIGCOMM, ACM IMC, ACM CoNEXT, PAM, IEEE INFOCOM, among others.

1 0

Jonathan Frankle will present his Masters talk on Friday, July 17, 2015 at 10am in rm 401.
by Nicki Gotsis 16 Jul '15

16 Jul '15

Jonathan Frankle will present his Masters talk on Friday, July 17, 2015 at 10am in rm 401. The members of his committee are David Walker and Aarti Gupta. Everyone is invited to attend his talk. Title and abstract are below. Example-Directed Synthesis: A Type-Theoretic Interpretation Input-output examples have emerged as a practical and user-friendly specification mechanism for program synthesis in many environments. While example-driven tools have demonstrated tangible impact that has inspired adoption in industry, their underlying semantics are less well understood: what are "examples" and how do they relate to other kinds of specifications? I describe examples as "refinement types." Seen in this light, program synthesis is the task of finding an inhabitant of such a type. This insight provides an immediate semantic interpretation for examples. Moreover, in makes it possible to exploit decades of research in type theory as well as its correspondence with intuitionistic logic rather than designing ad hoc synthesis systems from scratch.

1 0

Tianqiang Liu will present his FPO Thursday, 7/23/2015 at 2pm, 401 Computer Science Building. "Analyzing, Optimizing and Synthesizing Scenes by Reasoning About Relationships Between Objects"
by Nicki Gotsis 16 Jul '15

16 Jul '15

Tianqiang Liu will present his FPO, "Analyzing, Optimizing and Synthesizing Scenes by Reasoning About Relationships Between Objects", Thursday, 7/23/2015 at 2:00pm, 401 Computer Science Building. The members of his committee are Thomas Funkhouser (adviser); Readers: Szymon Rusinkiewicz, Wilmot Li (Adobe Research); Non-readers: David Dobkin, Adam Finkelstein A copy of his thesis, is available in Room 310. Everyone is invited to attend his talk.

1 0

Nanxi Kang will present her Pre-FPO on Monday, June 29 at 12pm in CS 401.
by Nicki Gotsis 22 Jun '15

22 Jun '15

Nanxi Kang will present her Pre-FPO on Monday, June 29 at 12pm in CS 401. The members of her committee are: Jennifer Rexford (adviser), David Walker, Sanjay Rao (Purdue), Nick Feamster, Mike Freedman. Everyone is invited to attend her talk. The the abstract follow below: Managing large enterprise networks is challenging. Operators need diverse policies such as routing, access control, QoS and load balancing, for performance, security and reliability. The best practice to enforce these policies today is to configure rules on network switches individually. This forces operators to reason about many low-level details, all at the same time, including the choice of path, the rule-table space limits on each switch, and the hop-by-hop interaction of rules for forwarding, dropping, modifying, and monitoring packets. The rule-table space, in particular, is a scarce commodity on switches. While the rule-tables (e.g., TCAM) are optimized for high-speed arbitrary packet-header matching and widely used for ACL, QoS and forwarding, they have small capacities on the order of a few thousand entries. In this talk, we propose a series of abstractions that enable operators to specify high-level goals without worry about the configurations of underlying switches, and algorithms that realize these abstractions within the constrained rule-table sizes. We first present the "One-Big-Switch" abstraction, which virtualizes the network topology as a single switch and allows operators to specify network policies as if configuring a single switch. The rule placement algorithm is responsible for distributing the rules for this single switch to the underlying physical switches. With this abstraction, we also develop a traffic monitoring system, PacketSeer, which acts as "Wireshark" running on top of the "One-Big-Switch" to aid measurement and troubleshooting. The second abstraction is "One-Big-Server", which consists of a set of equivalent servers offering the same services (e.g., websites). Given the desired distribution of server loads, Niagara algorithms compute forwarding rules on a switch to split incoming traffic accurately. Finally, we propose "Attribute-Carrying IPs". ACIP allows operators to define policies for groups of hosts with the same attributes (e.g., owner role or department). Our ACIP allocation algorithms effectively compact these policies defined on host attributes (e.g., faculty or students, CS or EE) into a small number of switch rules. Together, these abstractions -- coupled with efficient algorithms for realizing the abstractions -- enables flexible enterprise network management on commodity switches.

1 0