BEGIN:VCALENDAR METHOD:PUBLISH PRODID:-//Apple Inc.//iCal 3.0//EN CALSCALE:GREGORIAN X-WR-CALNAME:CMU_SCS-CS X-WR-RELCALID:F5EE02D9-77C2-4B80-8A1A-84F6CB057230 VERSION:2.0 X-WR-TIMEZONE:US/Eastern BEGIN:VTIMEZONE TZID:US/Eastern BEGIN:DAYLIGHT TZOFFSETFROM:-0500 TZOFFSETTO:-0400 DTSTART:20070311T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0400 TZOFFSETTO:-0500 DTSTART:20071104T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT SEQUENCE:3 TRANSP:OPAQUE UID:7E1AF4E1-F51C-463B-97E3-9BBF2EA6D090 DTSTART;TZID=US/Eastern:20081208T120000 DTSTAMP:20080929T193707Z SUMMARY:CS Faculty Lunch CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081208T130000 END:VEVENT BEGIN:VEVENT SEQUENCE:3 TRANSP:OPAQUE UID:5935830D-CCFD-492C-B91A-F1F04A5CF829 DTSTART;TZID=US/Eastern:20081013T120000 DTSTAMP:20080929T193559Z SUMMARY:CS Faculty Lunch CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081013T130000 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\nWe present a unified approach to type checking an d property checking for low-level code. Type checking for low-level code is challenging because type safety often depends on complex\, program-s pecific specific invariants that are difficult for traditional type chec kers to express. Conversely\, property checking for low-level code is ch allenging because it is difficult to write concise specifications that d istinguish between locations in an untyped programs heap. We address bot h problems simultaneously by implementing a type checker for low-level c ode as part of our property checker HAVOC. We present a low-level formal ization of a C program's heap and its types that can be checked with an SMT solver\, and we provide a decision procedure for checking type safet y. Our type system is flexible enough to support a combination of nomina l and structural subtyping for C\, on a per-structure basis. We discuss several case studies that demonstrate the ability of this tool to expres s and check complex type invariants in low-level C code\, including seve ral small Windows device drivers.\n \nThis is joint work with Jeremy Con dit and Shaz Qadeer (MSR Redmond)\, and intern Brian Hackett (Stanford)\ , and will appear in POPL09. The work is part of an ongoing effort of pe rforming precise and scalable analysis of low-level systems code in HAVO C. UID:C7D19B63-B414-4931-8703-988800D92E15 TRANSP:OPAQUE URL;VALUE=URI:message:%3C007901c93b78$37857fe0$42da0280@adm.cs.cmu.edu%3 E DTSTART;TZID=US/Eastern:20081031T140000 DTSTAMP:20081031T165524Z SUMMARY:SVC seminar — Unifying Type Checking and Property Checking for L ow-Level Code in HAVOC\, Shuvendu Lahiri - Microsoft Research CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20081031T150000 LOCATION:Redmond \, WA 98052 END:VEVENT BEGIN:VEVENT SEQUENCE:6 DESCRIPTION:Years 1-3 UID:A4C05BAA-9C52-491D-BE18-2C1B3D0DF583 TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20081219T100000 DTSTAMP:20081001T144738Z SUMMARY:Black Friday - General Meeting CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20081219T160000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:THE TURING AWARD LECTURE (a.k.a. ED11111010 !)\nThursday\, 23 October 2008\n-----3:30 pm - Extremely Distinguished Donuts (outside the auditorium)\n-----4:00 pm - The Turing Award Lecture\nReception in h onor of our Speaker will follow in the Perlis Atrium\, Newell-Simon Hall \nThe community\, alumni and visitors are welcomed.\n\nEDMUND M. CLARKE\ nFORE Systems University Professor of Computer Science\nCarnegie Mellon University\nand 2007 Turing Award Recipient\n\nMODEL CHECKING: \nMY 27- YEAR QUEST TO OVERCOME THE STATE EXPLOSION PROBLEM\n\nModel Checking is an automatic verification technique for state-transition systems that \n are finite-state or that have finite-state abstractions. In the early 19 80’s in a series of joint \npapers with my graduate students E.A. Emerso n and A.P. Sistla\, we proposed that \nModel Checking could be used for verifying concurrent systems and gave algorithms \nfor this purpose. At roughly the same time\, Joseph Sifakis and his student J.P. Queille \nat the University of Grenoble independently developed a similar technique. \nModel Checking has been used successfully to reason about computer ha rdware and \ncommunication protocols and is beginning to be used for ver ifying computer software. \nBy expressing a system’s specifications in t emporal logic\, the Model Checker can \nperform an exhaustive search to determine if the specification is true. In those cases \nwhere the speci fication does not hold\, the Model Checker produces a counterexample \ne xecution trace. We have found this feature to be extremely useful for fi nding obscure \nerrors in complex systems. However\, Model Checking is v ulnerable to the state-explosion \nproblem\, which can occur if the syst em being verified has many processes that execute \nin parallel or compl ex data structures. In some cases\, the state-explosion problem is \nine vitable\, but over the past 27 years we have made considerable progress in overcoming \nthis problem for certain classes of state-transition sys tems that occur often in practice. \nn this talk\, I will describe what Model Checking is\, how it works\, and the main techniques\n that have b een developed for combating the state-explosion problem.\n UID:E4785950-79A2-490D-A4B0-427202D56144 TRANSP:OPAQUE URL;VALUE=URI:message:%3C48F0B4B5.8090806@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081023T153000 DTSTAMP:20081017T141831Z SUMMARY:SCS DISTINGUISHED HOMECOMING LECTURE: EDMUND CLARKE\, The Turin g Award Lecture! CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20081023T173000 LOCATION:Wean 7500 and Perlis Atrium END:VEVENT BEGIN:VEVENT SEQUENCE:4 DESCRIPTION:Actual Mondays\, possible move to 12:30 and frequency to be decided at this meeting. UID:868C0A09-D577-459D-85AA-D0666A69C460 TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20080929T120000 DTSTAMP:20080925T161354Z SUMMARY:CS Faculty Lunch CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20080929T130000 LOCATION:Wean 7220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nThe search problem is a fundamental problem in theoretical computer science that consists of locating the region that c ontains a query point (or serving a sequence of such queries). In many c ases of this problem\, such as 1D binary search and planar point locatio n\, optimal algorithms are known when worst-case input is assumed. Howev er\, if we are concerned with speeding up the performance of a sequence of queries in a practical application\, the sequence of queries is often highly nonrandom. For example\, each successive query might be near to a recently queried element. This motivates the search for algorithms tha t have a provably guaranteed speedup for some of the kinds of nonrandomn ess a search algorithm is likely to encounter. This thesis proposal will cover work toward this end\, discussing lower bounds for the binary sea rch tree model\, both adaptive and competitive algorithms in the binary search tree model\, and adaptive search in dimensions higher than 1.\n\n Thesis Committee:\nDaniel Sleator\, Chair\nGary Miller\nGuy Blelloch\nSe th Pettie\, University of Michigan\n\nThesis Summary: http://www.cs.cmu. edu/~jonderry/proposal.pdf\n UID:E0F2345F-D6FC-48D3-9215-19BFD364831B TRANSP:OPAQUE URL;VALUE=URI:message:%3C4921EE2D.1040002@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081124T133000 DTSTAMP:20081118T160508Z SUMMARY:Thesis Proposal - Adaptive Search Data Structures\, Jonathan C. Derryberry CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20081124T143000 LOCATION:8220 Wean Hall END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n LEO-II is a standalone\, resolution-based highe r-order theorem\n prover designed for effective cooperation with specia list\n provers for natural fragments of higher-order logic.\n At prese nt LEO-II can cooperate with the first-order automated\n theorem prover s E\, SPASS\, and Vampire. LEO-II is implemented in\n Objective Caml an d its problem representation language is the new\n TPTP THF language.\n We demonstrate that problems in normal multimodal logic and\n access control logics can be elegantly encoded simple type theory\n and automa ted in LEO-II.PAL Colloquium Page:\n/www.andrew.cmu.edu/user/awodey/coll oquium/talks.html\n UID:DE0ED0C8-00DA-4578-B8DA-0382455B6C6C TRANSP:OPAQUE URL;VALUE=URI:message:%3C8eef88840811171104o2193e45cuffa493c88dc69371@ma il.gmail.com%3E DTSTART;TZID=US/Eastern:20081119T163000 DTSTAMP:20081118T160712Z SUMMARY:PAL Colloquium: Automating Access Control Logics and Multimodal Logics in the Automatic Higher-Order Theorem Prover LEO-II\, Christoph B enzmuller\, Saarland University CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20081119T173000 LOCATION:Baker Hall 237B END:VEVENT BEGIN:VEVENT SEQUENCE:6 DESCRIPTION:Years 4-N UID:4BFEAE7F-7022-4D16-8C38-D86DB86F386A TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20081218T130000 DTSTAMP:20081001T144658Z SUMMARY:Black Friday - Systems CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081218T160000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:7CAB1C92-F25C-4316-82C1-3C09D4DF32A3 DTSTART;TZID=US/Eastern:20081027T120000 DTSTAMP:20081020T201542Z SUMMARY:CS Faculty Lunch CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081027T130000 END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:ABSTRACT:\nCausal request traces are valuable to developers of large concurrent and\ndistributed applications\, yet difficult to obt ain. Traces show how a request is processed\, and can be analyzed by to ols to detect performance or correctness errors and anomalous behavior.\ n\nBorderPatrol obtains precise request traces through systems built fro m a litany of unmodified modules. Traced components include Apache\, th ttpd\,\nPostgreSQL\, TurboGears\, BIND and notably Zeus\, a closed-sourc e event-driven web server. BorderPatrol obtains traces using active obse rvation which\ncarefully modifies the event stream observed by modules\, simplifying precise observation. Protocol processors leverage knowledge about standard\nprotocols\, avoiding application-specific instrumentati on. Using these techniques BorderPatrol obtains precise traces for blac k-box systems that\ncannot be traced by any other technique.\n\nBIO:\nJo hn Jannotti is an Assistant Professor of Computer Science at Brown\nUniv ersity. At Brown\, John's research includes simplifying the design of\n large-scale distributed systems\, accountable yet anonymous peer-to-peer \nsystems\, and smart camera networks. Before joining Brown\, John rece ived his\nPhD from the Massachusetts Institute of Technology. While at MIT\, John\nworked on operating system architectures\, software routers\ , wireless\nprotocols\, and next-generation networking designs.\n UID:81CB3380-C06E-4772-8EF1-491285CDA281 TRANSP:OPAQUE URL;VALUE=URI:message:%3C008c01c9312f$24946a20$66860280@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20081023T120000 DTSTAMP:20081020T141025Z SUMMARY:SDI / LCS Seminar - Precise Causal Traces of Black-box Applicati ons with BorderPatrol\, John Jannotti\, Brown University CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081023T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:We are pleased to announce Sharad Agarwal\, Researcher in th e Networking Group at Microsoft Research\, will be our guest speaker at the CyLab Seminar\, on Wednesday\, December 17h at 12 pm Eastern. He wil l be presenting "Bach: A Network Coordinate System that Harnesses the In ternet’s Topology."\n\nPlease join us in Room 2101 (*please note the roo m change) on the second Floor of the CIC building. Lunch will be provid ed. UID:C640BD22-B0D7-4CDD-ADE5-224EA4A3467E TRANSP:OPAQUE URL;VALUE=URI:message:%3C8B2D828C031E483A8490177DC6134292@andrew.ad.cmu. edu%3E DTSTART;TZID=US/Eastern:20081219T170000 DTSTAMP:20081119T194011Z SUMMARY:CyLab Seminar - Bach: A Network Coordinate System the harnesses the Internets Topology\, Sharad Agarwal\, Researcher in the Networking G roup at Microsoft Research CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20081219T180000 LOCATION:CIC 2101 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nRAID systems are under pressure to reconstruct da ta faster because of ever\nlarger disk capacities. The parity computati on is the core function of\nRAID controllers. Although traditionally im plemented in a dedicated ASIC\ndevice\, more and more storage vendors ar e using commodity processor\nresources to compute RAID.\n\nThis talk loo ks at using graphics co-processor resources becoming\navailable for high performance computing\, specifically the NVIDIA GeForce\n880 GTX CUDA\, for implementing RAID without special purpose ASIC support.\n\nBIO:\nLe e Ward is a principal member of technical staff at Sandia National\nLabo ratories. As an inveterate student of operating systems and file\nsystem s\, his interests have provided the opportunity to make contributions\ni n high performance\, parallel file systems\, IO libraries\, hierarchical \nstorage management\, and compute cluster integration/management system s. \nThe IO performance of Sandia current largest computer was recently\ ndocumented by his team:\npapers.cluster2007.org/presentations/RedStormP erformanceAnalysis.pdf UID:1496E470-58B7-40DA-A3D6-0E862402EE15 TRANSP:OPAQUE URL;VALUE=URI:message:%3C00eb01c91f1f$ea698180$66860280@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20080925T120000 DTSTAMP:20080925T154251Z SUMMARY:SDI/LCS Seminar Efficacy of GPUs in RAID Parity Calculations\, Lee Ward\, Sandia National Laboratory CREATED:20081125T171250Z DTEND;TZID=US/Eastern:20080925T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nThe throughput of existing wireless networks is o ften limited by\ninterference. One fundamental reason is that the curren t designs are\nconstrained by a "one-transmission-at-a-time" model at th e link layer and a\nfixed-width spectrum allocation at the physical laye r. We present a new\nwireless design that exploits traffic burstiness an d node heterogeneity\,\nthereby improving concurrency and spectrum usage . The main challenge is the\nunmanaged nature of many wireless networks such as 802.11 and mesh\, which\nmakes centralized resource allocation i mpractical. We show through analysis\nand implementation that simple ran domized allocation policies can overcome\nthis challenge\, and improve t hroughput by 2x or more.\n\nThis work is joint with Rabin Patra\, Hari B alakrishnan and Eric Brewer. UID:5E6F65FA-63FB-4359-ADCD-CC180EA5CCC5 TRANSP:OPAQUE URL;VALUE=URI:message:%3C9C70373506B54383AA2E5890A6CB4940@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20081120T120000 DTSTAMP:20081118T145602Z SUMMARY:SDI / LCS Seminar - Interference rendered significantly harmless \, Ramakrishna Gummadi\, MIT - CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081120T130000 LOCATION:CIC 2101 END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:605F6A27-BC4C-4A4B-A4B5-F038DA0655D3 DTSTART;TZID=US/Eastern:20081029T103000 DTSTAMP:20081029T135003Z SUMMARY:Parallelism for the Masses: Opportunities and Challenges\, Andre w A. Chien\, Vice President of Research\, Intel Corporation CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081029T113000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:3 TRANSP:OPAQUE UID:0FEA77DC-BD46-4301-9E1D-5F898987D0C7 DTSTART;TZID=US/Eastern:20081110T120000 DTSTAMP:20080929T193644Z SUMMARY:CS Faculty Lunch CREATED:20081125T171250Z DTEND;TZID=US/Eastern:20081110T130000 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:C25760E4-33FE-44BE-9235-A987A2B86686 DTSTART;TZID=US/Eastern:20081017T120000 URL;VALUE=URI:message:%3C48ECD494.1000603@cs.cmu.edu%3E DTSTAMP:20081008T163348Z SUMMARY:Speaking Skills Talk - A Tale of Two Consequence Relations\, Dan Licata CREATED:20081125T171252Z DTEND;TZID=US/Eastern:20081017T130000 LOCATION:WeH 5409 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nBursts in data center workloads are a real proble m for storage subsystems. Data volumes can experience peak I/O request r ates that are over an order of\nmagnitude higher than average load. This requires significant\nover-provisioning\, and often still results in si gnificant I/O request\nlatency during peaks.\n\nIn this talk I'll presen t Everest\, an architecture that allows data written\nto an overloaded v olume to be temporarily off-loaded into a short-term virtual store. Ever est creates the short-term store by opportunistically pooling underutili zed storage resources either on a server or across servers\nwithin the d ata center. Writes are temporarily off-loaded from overloaded volumes to lightly loaded volumes\, thereby reducing the I/O rate on the former. W e evaluated Everest using traces from a production Exchange mail server as well as other benchmarks: our results show a 1.4--70 times\nreduction in mean response times during peaks.\n\nThis is joint work with my coll eagues at MSR Cambridge: Austin Donnelly\, Sameh Elnikety\, Eno Thereska \, and Ant Rowstron. An early prototype of write\noff-loading was used t o save energy by increasing disk idle time\, and was presented at FAST 2 008. The Everest work will be presented at OSDI 2008. UID:70E4FCB2-4D72-4C60-B1B6-5F9A02274CF3 TRANSP:OPAQUE URL;VALUE=URI:message:%3C000f01c93aa2$d08bb6e0$66860280@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20081030T120000 DTSTAMP:20081030T153637Z SUMMARY:SDI/LCS Seminar - Everest: scaling down I/O peaks through write off-loading\, Dushyanth Narayanan\, Microsoft Research\, Cambridge CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081030T130000 LOCATION:CIC END:VEVENT BEGIN:VEVENT SEQUENCE:6 DESCRIPTION:Years 4-N UID:40B060B5-920C-49A7-8A47-8741E28A2D90 TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20081218T093000 DTSTAMP:20081001T144624Z SUMMARY:Black Friday -AI & Theory CREATED:20081125T171251Z DTEND;TZID=US/Eastern:20081218T120000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:TRANSPARENT UID:016F36B6-7679-4632-A7D9-5D77687EC7C4 DTSTART;VALUE=DATE:20090226 DTSTAMP:20081013T172515Z SUMMARY:CSD Open House CREATED:20081125T171250Z DTEND;VALUE=DATE:20090302 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nObject recognition is a fundamental capability for the potential robot assistance to humans in useful tasks. There have been numerous visual-based object recognition systems\, fast and accura te results in constrained environments. However\, by depending on visual cues\, these techniques are susceptible to object views variations in s ize\, lighting\, rotation\, and pose\, all of which cannot be avoided in real visual data. Thus\, it is widely acknowledged that the general obj ect recognition task still remains very challenging.\n\nMy thesis work b uilds upon the fact that robots can observe humans interacting with the objects in their environment\, and that humans provide numerous non-visu al cues to the identity of objects. A few previous frameworks have succe ssfully incorporated a visual and some non-visual cue\, such as activity . However\, all of the methods are limited to a specific additional cue\ , and where the type of the cue must be specified at the outset of the f ramework.\n\nIn this thesis\, I will introduce a flexible object recogni tion approach which can use multiple cues of any predefined type\, wheth er they are cues intrinsic to the object or provided by observation of a human. In pursuit of this goal\, the thesis will provide several contri butions: A representation for the multiple cues including an object defi nition that allows for the flexible addition of these cues\; Weights tha t reflect the various strength of association between a particular cue a nd a particular object using a probabilistic relational model and\; Last ly\, an algorithm for the incremental discrimination of the object set f or recognition and a framework for adapting these definitions to specifi c contexts while learning general properties. I will be evaluating this algorithm with real video data and an extensive set of objects and envir onments.\n\nThesis Committee:\nManuela Veloso\, Chair\nMartial Hebert\nA lexei Efros\nIrfan Essa\, Georgia Institute of Technology\n\nThesis Summ ary: http://www.cs.cmu.edu/afs/cs/user/saboutal/www/proposal.pdf\n\n UID:C9B6B480-BB82-4502-8445-80F5F02CB2AF TRANSP:OPAQUE URL;VALUE=URI:message:%3C492D6894.4060803@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081205T093000 DTSTAMP:20081126T154829Z SUMMARY:Thesis Proposal - Context-Dependent Multi-Cue Object Recognition \, Sarah Aboutalib CREATED:20081126T154829Z DTEND;TZID=US/Eastern:20081205T103000 LOCATION:3002 Newell-Simon Hall END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nConventional secure desktop solutions running o n top of the OS have provided user-friendly\, per-application sandboxes for executing untrusted code\, but they have also suffered from limited granularity and accuracy in monitoring such code. And while other sandbo xing techniques such as simple full-system virtualization have been used to safely execute and monitor untrusted code\, they lacked the scalabil ity of the monitoring system itself.\n\nWe introduce VMM Sandbox\, a vir tualized sandbox environment that is lightweight\, fast\, scalable\, and hence suitable for use in both program monitoring and secure desktop so lutions. Based on the Xen VMM\, it uses flash cloning techniques to "for k" virtual machines per application instance for scalability. It also pr ovides a GUI package for driving the system as a secure desktop solution . We also propose a lightweight communication scheme and a device driver for the guest virtual machine that will further improve the performance of the system.\n\nThesis Committee:Â\nJames C. Hoe\, Chair\nHyong S. Ki m\nAdrian Perrig\n\nThesis summary: http://www.andrew.cmu.edu/~yongjunj/ thesis.pdf\n UID:4144A813-9656-49D6-9542-213BC8D430E0 TRANSP:OPAQUE URL;VALUE=URI:message:%3C492D7EC6.2070607@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081203T160000 DTSTAMP:20081126T171625Z SUMMARY:Masters Thesis Presentation - VMM Sandbox\, Yongjun Jeon CREATED:20081126T171625Z DTEND;TZID=US/Eastern:20081203T170000 LOCATION:4625 Wean Hall END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nAs people increasingly use mobile devices and Int ernet services to manage and share large data collections\, exploiting c ommunication proximity and selectively replicating content is essential. Cimbiosys is a replicated storage platform that permits each device to define its own content-based filtering criteria and to share updates di rectly with other devices. This talk focuses on the key challenges in p roviding content-based partial replication in the face of fluid network connectivity\, redefinable content filters\, and changing content. I wi ll discuss the design of Cimbiosys \, which ensures two fundamental prop erties. First\, every device eventually stores precisely those items wh ose latest version matches its custom filter. Second\, every device rep resents its replication-specific metadata in a compact form\, resulting in low data synchronization overhead. This permits ad hoc replication b etween newly encountered devices with overlapping interests and frequent synchronization between established partners\, even over low bandwidth wireless networks.\n \nBIO:\nDoug Terry is a Principal Researcher in the Microsoft Research Silicon Valley lab. His research focuses on the des ign and implementation of novel distributed systems and addresses issues such as information management\, fault-tolerance\, and mobility. He cu rrently is serving as Chair of ACM's Special Interest Group on Operating Systems (SIGOPS). Prior to joining Microsoft\, Doug was the co-founder and CTO of a start-up company named Cogenia\, Chief Scientist of the Co mputer Science Laboratory at Xerox PARC\, and an Adjunct Professor in th e Computer Science Division at U. C. Berkeley\, where he still occasiona lly teaches a graduate course on distributed systems. Doug has a Ph.D. in Computer Science from U. C. Berkeley.\n \nVisitor Host: Satya\nVisito r Coordinator: Tracy Farbacher\, tracyf@cs.cmu.edu\n \nSDI/LCS Seminar q uestions: Karen Lindenfelser\, karen@ece.cmu.edu UID:49A0112F-4A68-4E09-A3FF-FD1B2A25AF32 TRANSP:OPAQUE URL;VALUE=URI:message:%3C7F8AA29D857B4234896AEB69470FB9B7@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20081211T120000 DTSTAMP:20081202T181826Z SUMMARY:SDI / LCS Seminar - Content-based Data Replication: Challenges a nd Techniques — Speaker: Doug Terry\, Microsoft Research Silicon Valley CREATED:20081202T181826Z DTEND;TZID=US/Eastern:20081211T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nA"k-CSP" is a system of constraints over n bool ean-valued variables v_i\nin which each constraint involves at most k of the variables. Given a\nk-CSP instance\, Max k-CSP is the algorithmic p roblem of finding an\nassignment for the variables that satisfies the ma ximum fraction of the\nconstraints. A k-CSP instance is called "satisfia ble" if there is such\nan assignment that satisfies all the constraints. \n\nFor a satisfiable 3-CSP instance\, it is NP-hard to find the optimal \nassignment\; however there is a 5/8-approximation algorithm for the\np roblem by Zwick [zwi98]. In this talk we will review Zwick's algorithm\n and show why it is NP-hard to have a better approximation algorithm\nass uming some conjecture.\n\nPresented in Partial Fulfillment of the CSD Sp eaking Skills Requirement.\n\n UID:A02522FB-1BD6-4411-8A0C-C399BCABE8CE TRANSP:OPAQUE URL;VALUE=URI:message:%3C493FD2F9.4050007@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081210T120000 DTSTAMP:20081210T144438Z SUMMARY:Speaking Skills Talk - Hardness of Approximating Satisfiable 3CS Ps\, Yi Wu CREATED:20081210T144438Z DTEND;TZID=US/Eastern:20081210T130000 LOCATION:1507 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nI will talk about a technique for recognizing n onrigid objects under length preserving deformations. Such deformations are pervasive\, e.g.\, varying facial expressions\, bending limbs or par ts of living beings? bodies\, and moving components of articulated objec ts. The technique consists of three stages: first measuring the geodesic distances between uniformly distributed points on the 3D surface of the object\; then transforming the object in a finite dimensional Euclidean space in which geodesic distances are replaced by Euclidean ones\; and matching different objects in the space as they are rigid. This techniqu e was originally developed by Asi Elad and Ron Kimmel from Israel Instit ute of Technology\, and applied to face and human body recognition by va rious researchers.\n\nPresented in Partial Fulfillment of the CSD Speaki ng Skills Requirement.\n UID:06DD4D5A-0FB2-45CB-92A7-6E2751C27EBD TRANSP:OPAQUE URL;VALUE=URI:message:%3C493FE339.1040201@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081216T120000 DTSTAMP:20081210T154255Z SUMMARY:Speaking Skills Talk - Bending Invariant Object Recognition\, Le on Gu CREATED:20081210T154255Z DTEND;TZID=US/Eastern:20081216T130000 LOCATION:Leon END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:Abstract:\n\nTraditional visual neuroscience research has fo cused on determining the relationship between the activity of single neu rons and the stimuli from the outside world\, and more recently the inte ractions within pairs of neurons. These studies have typically recorded from neurons or pairs of cells in isolation. Recent advances in neural r ecording devices have made it possible to record simultaneously from hun dreds of cells. Such data provide new insights into the interactions amo ng the neurons\, the connectivity of neurons in a local network\, as wel l as the neural algorithms of information processing. These methods also present new challenges: the scaling of existing system identification a nd decoding techniques to address the dramatic increase in dimensionalit y and computational complexity\, and the development of new statistical methods to infer the dynamic interaction and connectivity in neuronal en sembles during information processing.\n\nWe have recorded neuronal acti vity from the primate primary visual cortex using 96-channel micro-elect rode arrays during the presentation of a variety of visual stimuli. I wi ll talk about modeling this data using the generalized linear model fram ework to explain and predict the spike train of each recorded neuron as a function of the input stimulus\, the recorded spiking activity of near by neurons\, as well as the local field potentials recorded across the m icro-electrode array.\n\nPresented in Partial Fulfillment of the CSD Spe aking Skills Requirement. UID:9AE087F1-DA0A-4200-BBF4-EABF382DC48C TRANSP:OPAQUE URL;VALUE=URI:message:%3C493FE604.2080504@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081217T120000 DTSTAMP:20081210T155516Z SUMMARY:Speaking Skills Talk - Analysis of large-scale neuronal data wit h the GLM\, Ryan Christopher Kelly CREATED:20081210T155502Z DTEND;TZID=US/Eastern:20081217T130000 LOCATION:4623 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:While techniques to mask identifying information from transm issions\nat and above the bit level are essential for strong privacy\npr otection\, mobile systems research frequently ignores that\nsensitive in formation can also be obtained at the wireless signal\nlevel.\n\nIn this talk\, we will highlight our recent privacy results---both breaches and protections---from the signal domain. Specifically\,\nwhile it is widel y known that it is possible to localize\ntransmissions at the signal lev el\, we will show that one can\nalso identify transmitters with surprisi ng reliabilty---even when\nignoring message content.\n\nIt is also feasi ble to derive higher-order knowledge from signals\,\nfor example which d evices have been co-located\, without obtaining\ndevice locations. Fortu nately\, some privacy-enhancing mechanisms do\nexist. Path cloaking algo rithms can be adapted for regulating\ntransmission timing to prevent lon g-term movement tracking of\ntransmitters. We will also discuss prelimin ary results that\ndemonstrate thwarting of precise localization through cooperation\namong transmitters. These results show that much work remai ns in\nincorporating signal level privacy in mobile systems designs.\n\n -- \nJason I. Hong\, Asst. Prof. http://www.cs.cmu.edu/~jas onh\nHCI Institute\, School of Computer Science\, Carnegie Mellon Univer sity\n UID:CCA834D1-96C3-4C66-AE58-AB73353F70C3 TRANSP:OPAQUE URL;VALUE=URI:message:%3C49402810.700@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20081211T150000 DTSTAMP:20081210T205334Z SUMMARY:Special Talk: Reading Signals: Radiometric Breaches and Enhancem ents of\nWireless Location Privacy\, Marco Gruteser from Rutgers CREATED:20081210T205334Z DTEND;TZID=US/Eastern:20081211T160000 LOCATION:NSH 1507 END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:SPEAKER:\nSharad Agarwal - Networking Group\, Microsoft Rese arch\nhttp://research.microsoft.com/~sagarwal/\n\nTITLE:\nBach: A Networ k Coordinate System that Harnesses the Internet's Topology\n\nABSTRACT:\ nPredicting network latency is important for several distributed systems .\nNetwork coordinate systems (NCSs) are a popular approach to latency\n prediction because they are decentralized\, scalable\, inclusive and ada ptive.\nHowever\, NCSs use an abstract world model divorced from the und erlying\nnetwork topology\, resulting in systematic errors and slow conv ergence. We\ndemonstrate dramatic improvements in performance without sa crificing the\nproperties prized in an NCS\, by incorporating aspects of the Internet's\ntopology to better ground its model in reality. We pre sent four simple\ntechniques for improving an NCS in this way: geographi c bootstrapping\,\nautonomous system correction\, history prioritization \, and symmetric updates.\nWe illustrate the effectiveness of these appr oaches by building a system\,\nBach\, that incorporates them all\, and b y evaluating each aspect. Our\nevaluation uses both a real-world deploy ment and a replay of far-reaching\ntraces\, covering over 4.9 million ma chines and an enormous number\, 91\nmillion\, of measurements between th em. We find that Bach has substantially\nbetter predictive power for ho me machines than current state-of-the-art\nsystems\, including Vivaldi.\ n\nBIO:\nSharad Agarwal is a researcher in the Networking Group at Micro soft\nResearch. He is also affiliated with the Systems and Networking Gr oup. He\ncompleted his Ph.D. in September 2004 in Computer Science at UC Berkeley.\nHis work was in wide-area routing\, traffic performance and network\nmanagement under the supervision of Professor Randy Katz. Durin g his Ph.D.\nhe spent two years with the IP and Security Research group at Sprint Labs.\nHe received his M.S. in December 2000\, also in Compute r Science at UC\nBerkeley. During his M.S.\, he spent time at HRL Labs. Being an avid Bears\nfan\, he also has a B.S. in Electrical Engineering and Computer Science from\nUC Berkeley.\n\nPlease direct any questions r egarding this talk to Sihyung Lee\n(sihyunglee@cmu.edu). UID:8E6F3827-4527-4390-B599-5BB79AC7C66B TRANSP:OPAQUE URL;VALUE=URI:message:%3CCBBA7F7CD23342F08FC949FF7C4BE972@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20081217T120000 DTSTAMP:20081217T142829Z SUMMARY:CyLab - Special Seminar - Bach: A Network Coordinate System that Harnesses the Internet's Topology\, Sharad Agarwal\, Microsoft Research CREATED:20081210T215608Z DTEND;TZID=US/Eastern:20081217T130000 LOCATION:CIC 2101 END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:0AFBDBB8-1183-4581-B1A5-30D4B4C04EFC DTSTART;TZID=US/Eastern:20081223T150000 DTSTAMP:20081216T175345Z SUMMARY:Help Desk closed for holidays CREATED:20081216T175218Z DTEND;TZID=US/Eastern:20081223T235900 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:TRANSPARENT UID:E2BF341D-259F-4588-BB00-79A784A6AA9D DTSTART;VALUE=DATE:20081224 DTSTAMP:20081216T175401Z SUMMARY:Help Desk Closed for Holidays CREATED:20081216T175301Z DTEND;VALUE=DATE:20090102 END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:01643000-5C82-4D79-A624-AAAC6A3D776D DTSTART;TZID=US/Eastern:20090102T083000 DTSTAMP:20081216T175455Z SUMMARY:Main Office re-opens after holiday CREATED:20081216T175405Z DTEND;TZID=US/Eastern:20090102T093000 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:478211E7-E779-4D01-870F-C03873F42B05 DTSTART;TZID=US/Eastern:20090102T090000 DTSTAMP:20081216T175437Z SUMMARY:Help Desk re-opens after holiday CREATED:20081216T175405Z DTEND;TZID=US/Eastern:20090102T100000 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:TRANSPARENT UID:8793CE4E-BBA5-41FC-97CC-F910680D71D1 DTSTART;VALUE=DATE:20081224 DTSTAMP:20081216T175522Z SUMMARY:Main Office closed for holidays CREATED:20081216T175503Z DTEND;VALUE=DATE:20090102 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:AD368522-AA21-4E69-8F98-8355626F1AD1 DTSTART;TZID=US/Eastern:20090112T120000 URL;VALUE=URI:message:%3C5DD2A99DF4A04FCC9C4494D7C7E7304B@andrew.ad.cmu. edu%3E DTSTAMP:20090112T153653Z SUMMARY:CyLab Seminar: Dr Rosario Gennaro\, researcher at IBM\, Identity -Based Key Exchange Protocols CREATED:20090112T153653Z DTEND;TZID=US/Eastern:20090112T130000 LOCATION:INI Distributed Education Center (DEC) on the Lobby level of th e CIC building END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nIt is expected that in future battlefield and d isaster response situations\, teams of unmanned vehicles\, remote sensor s\, and people will work together\, using an ad hoc wireless network to communicate. Unmanned aerial vehicles (UAVs) acting as mobile communicat ion nodes have been proposed to supplement the ground-based network\, wh ich may be limited due to a noisy environment with obstructions and inte rference. In this talk I will describe graph theoretic and network flow problems of where to position the UAVs in order to facilitate communicat ion within groups of nodes\, and present several techniques for solving these problems.\n\nPresented in Partial Fulfillment of the CSD Speaking Skills requirement. UID:5C3718B1-D11C-41E6-A924-90ECA5F05310 TRANSP:OPAQUE URL;VALUE=URI:message:%3C49761E74.9090805@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090205T120000 DTSTAMP:20090120T190327Z SUMMARY:Speaking Skills Talk - Network Design Problems for Outdoor Multi -Agent Systems\, Steven Okamoto CREATED:20090120T190327Z DTEND;TZID=US/Eastern:20090205T130000 LOCATION:Wean 5409 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Everyday users are increasingly at the front lines of defens e against cyber security threats. Usable network security approaches thi s “user problem” by adapting technical systems to users in an effort to make protections as easy and as effective as possible to use. Dr. LaRose and his colleagues at Michigan State University have been working on a different aspect of the problem: adapting users to technical systems by motivating them to adopt specific existing precautions and to develop vi gilance against a wide range of emerging threats. The cyber-security pro blem is redefined as one of safety promotion and draws on parallels to t he problems encountered in health communication campaigns. Dr. LaRose wi ll examine relevant theories of health promotion as they relate to the c yber-security situation and review the results of NSF-funded research wh ich applied these concepts to online safety promotion. The potential of safety promotion to improve network security\, as well as the potential for well-intentioned efforts to misfire\, will be discussed. UID:BDB9EF71-3D8C-4F97-8B07-12C611ECD3B4 TRANSP:OPAQUE URL;VALUE=URI:message:%3Cba20b70c5ff1288c8788fcd9e7ffaaaf619.20090121155 802@mcsv11.net%3E DTSTART;TZID=US/Eastern:20090123T120000 DTSTAMP:20090121T160634Z SUMMARY:Joint ISR/CyLab/CUPS Seminar - January 16th\, Dr. Robert Larose\ , Promoting Online Safety: Health Communication Paradigms for Cyber- Se curity CREATED:20090121T160634Z DTEND;TZID=US/Eastern:20090123T130000 LOCATION:INI DEC in the CIC END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:3D880745-AAD1-49A2-B603-3A3B2229A66E DTSTART;TZID=US/Eastern:20090129T120000 URL;VALUE=URI:message:%3CB46A25F8CA2546C6A4A72FC3D6C7DB00@ece.cmu.edu%3E DTSTAMP:20090123T144258Z SUMMARY:SDI / LCS Seminar — Native Client - A Sandbox for Portable\, Unt rusted x86 Native Code\, Brad Chen\, Google CREATED:20090123T144258Z DTEND;TZID=US/Eastern:20090129T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:24E3435B-7270-4FE7-A9BA-39B432FFF4B5 DTSTART;TZID=US/Eastern:20090205T120000 URL;VALUE=URI:message:%3C559269D407E440E2933FF2E8878677E0@pdl.cmu.edu%3E DTSTAMP:20090130T184620Z SUMMARY:SDI / LCS Seminar - Xax: Leveraging Legacy Code to Deploy Deskto p Applications on the Web\, Jon Howell\, Microsoft CREATED:20090130T184620Z DTEND;TZID=US/Eastern:20090205T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:3 DESCRIPTION:Jane Street Capital is a proprietary trading company that ha s shifted from developing \nsoftware in mainstream programming languages to developing software almost entirely in \nOCaml\, a statically typed functional programming language that has only modest industrial\nuse. T he scope of the enterprise is small but growing: Jane Street now has ove r 30 OCaml \nprogrammers that have collectively written hundreds of thou sands of lines of OCaml code. \nOCaml is used for building everything fr om trading systems to research infrastructure to user \ninterfaces to sy stems administration tools. This talk will discuss the motivations behin d \nJane Street's adoption of OCaml\, and why we think that statically t yped functional \nprogramming languages are a good fit for the world of t rading and finance.\n\nBio:\nYaron Minsky obtained his PhD in Computer Sc ience from Cornell University in 2002 \nfocusing on distributed systems. In 2003\, he joined Jane Street Capital where he founded \nthe quantita tive research group there. He became a Managing Director of Jane Street in 2007.\n\n UID:0788F097-D98D-4497-90AB-68CDF3C65E41 TRANSP:OPAQUE URL;VALUE=URI:message:%3C498740D2.7070900@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090205T153000 DTSTAMP:20090202T185722Z SUMMARY:SCS DISTINGUISHED LECTURE: Caml Trading\, Yaron Minsky\, Jane S treet Capital CREATED:20090202T185440Z DTEND;TZID=US/Eastern:20090205T173000 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nSelf-adjusting computation is a technique that was born from my doctoral-thesis research here at CMU. It is motivated by a myriad of examples in a broad set of application domains including programming languages\, robotics\, scientific computing\, computational biology\, etc. The goal is to offer linguistic paradigms\, techniques\, and tools for developing safe and efficient software systems that can au tomatically respond to modifications to their data\, state\, or environm ent (due to a number of reasons\, e.g.\, user interactions\, changes in the physical world). To achieve this goal\, we develop techniques that combine ideas from type-systems\, programming-languages\, algorithms\, a nd system design in innovative ways.\n\nIn our work thus far\, we have p rimarily used high-level\, type-safe\, mostly functional languages\, e.g .\, SML\, to host self-adjusting-computation languages and systems. We recently started experimenting with ideas for supporting self-adjusting computation efficiently in the context of the ultimate low-level languag e: C. In this talk\, I will give a brief overview of self-adjusting com putation and describe our experience in developing the CEAL (read "seal" ) language: A C-based language for self-adjusting computation. UID:8688059B-5DD5-4C84-B6C7-F27A71EBE1DE TRANSP:OPAQUE URL;VALUE=URI:message:%3C498B16CA.90304@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090206T153000 DTSTAMP:20090205T184655Z SUMMARY:POP Seminar — Self-Adjusting Computation in C\, Umut Acar\, Toy ota Technological Institute and the University of Chicago CREATED:20090205T184655Z DTEND;TZID=US/Eastern:20090206T163000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT\n\nDatabases have the stigma of an association with (boring) enterprise\ndata management. The area of database research\, h owever\, has developed\na wide set of concepts and techniques with appli cability much beyond\ndepartments and employees.\n\nIn this talk\, I wil l show how the idea of declarative processing from\ndatabases can be app lied to computer games. I will describe our\njourney from declarative to imperative scripting languages for\ncomputer games\, and I will introdu ce the state-effect pattern\, a\ndesign pattern that enables game develo pers to design games that can\nbe programmed imperatively\, but processe d declaratively. I will then\nintroduce Scalable Games Language (SGL)\, our scripting language for\ngames\, and I will outline how database tech niques can be used to\nprocess SGL resulting in performance improvements \nof an order of magnitude or more compared to standard scripting langu ages.\n\nThis work is an instance of our general research direction wher e we\napply data-centric ideas to problems outside traditional data\nman agement targeting both scalability and ease of programming\, and I\nwill discuss some of the lessons that we learned. UID:9CD0C354-6099-4CA7-AA82-6157CAF933F5 TRANSP:OPAQUE URL;VALUE=URI:message:%3C498B391A.5020606@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090209T140000 DTSTAMP:20090205T194135Z SUMMARY:Declarative Processing for Computer Games\, Prof. Johannes Gehrk e\, Cornell CREATED:20090205T194135Z DTEND;TZID=US/Eastern:20090209T150000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nWhat do your wireless devices reveal about you? D o you think that WEP or \nWPA keeps you safe? Did you know that eavesdr oppers may still figure \nout where you live and other places that you'v e visited? How can you \nprotect your privacy in an increasingly wirele ss world?\n\nWireless capabilities are rapidly spreading beyond laptop c omputers to \neveryday consumer devices ranging from cell-phones and per sonal health \nmonitors to game controllers and digital cameras. This ev olving wireless \necosystem is increasingly pervasive and personal in it s usage\, and it \nheightens privacy risks that are already significant compared to wired \nnetworks. Unfortunately\, our understanding of the privacy risks \nassociated with low-level wireless protocols is limited. In this talk\, I \nshow that\, with nothing more than commodity hardwar e\, eavesdroppers can \ntrack and profile devices that use existing prot ocols such as 802.11 and \nBluetooth. Devices can be tracked even when the best known defenses are \nemployed\, such as temporary device addres ses and link layer encryption. \nThis is because protocol control inform ation\, which remains exposed\, can \nimplicitly identify devices even i f they do not transmit any explicit \nidentifiers. To improve privacy\, I demonstrate how to build efficient \nwireless protocols that do not re veal any identifying information.\n UID:43529896-9A93-4098-852E-B008232F67C9 TRANSP:OPAQUE URL;VALUE=URI:message:%3CAADDD0C4FD3142308AB3B7536DB14F9F@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20090219T120000 DTSTAMP:20090213T183009Z SUMMARY:SDI/LCS Seminar - Improving the Privacy of Wireless Protocols\, Jeffrey Pang\, CMU CREATED:20090213T183009Z DTEND;TZID=US/Eastern:20090219T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\nA central problem in concurrent programming is ho w to discipline programs so that different threads of control can only i nterfere in ways that are predictable and thus easy to reason about. In this talk\, we will discuss a type-based approach for statically enforci ng safe concurrency\, partially inspired by spatial and separation logic s. Our type system assigns to each runtime entity (resource) a dynamic spatial-behavioral type that specifies how such entity may be safely man ipulated by one or more threads of control. This contrasts with most oth er approaches that focus on identifying structural constraints to the co de that manipulates the resources. We will also show how flexible discip lines for control of interference\, able to deal with expressive imperat ive reference-passing higher-order programming languages\, such as objec t-oriented programming\, may be compositionally expressed in the framewo rk. UID:F8BA10E1-143D-49BF-8922-CEFC14CE3D6F TRANSP:OPAQUE URL;VALUE=URI:message:%3C499D66CD.6080009@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090225T153000 DTSTAMP:20090219T150723Z SUMMARY:POP Seminar — Dynamic Control of Interference with Spatial-Behav ioral Types\, Luís Caires CREATED:20090219T150723Z DTEND;TZID=US/Eastern:20090225T163000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:EA88907F-E4D7-4C3D-9AFE-CACEE8987B49 DTSTART;TZID=US/Eastern:20090223T120000 URL;VALUE=URI:message:%3C49A2B582.1020305@andrew.cmu.edu%3E DTSTAMP:20090223T145245Z SUMMARY:CyLab Seminar — Rebecca Herold "The Convergence of Information S ecurity\, Privacy\, and Compliance" CREATED:20090223T145245Z DTEND;TZID=US/Eastern:20090223T130000 LOCATION:INI Distributed Education Center (DEC) on the \nLobby level of the CIC building END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:7E39645D-13F0-4E5D-84B8-7548E1F6FB1D DTSTART;TZID=US/Eastern:20090225T163000 URL;VALUE=URI:message:%3C49A2F37A.70504@cs.cmu.edu%3E DTSTAMP:20090223T191450Z SUMMARY:ML/Google Seminar - Matching\, Embedding and Clustering for Grap hs and Data\, Tony Jebara CREATED:20090223T191450Z DTEND;TZID=US/Eastern:20090225T173000 LOCATION:Newell-Simon Hall 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nThe problem of learning a policy\, a task repre sentation mapping from world states to actions\, lies at the heart of ma ny robotic applications. One approach to acquiring a task policy is lear ning from demonstration\, an interactive technique in which a robot lear ns a policy based on example state to action mappings provided by a huma n teacher.\n\nThis thesis introduces Confidence-Based Autonomy\, a mixed -initiative single robot demonstration learning algorithm that enables t he robot and teacher to jointly control the learning process and selecti on of demonstration training data. This algorithm enables the robot to i dentify the need for and request demonstrations for specific parts of th e state space based on confidence thresholds characterizing the uncertai nty of the learned policy. The robot's demonstration requests are comple mented by the teacher's ability to provide supplementary corrective demo nstrations in error cases. An additional algorithmic component enables c hoices between multiple equally applicable actions to be represented exp licitly within the robot's policy through the creation of option classes .\n\nBased on the single-robot Confidence-Based Autonomy algorithm\, thi s thesis introduces a task and platform independent multi-robot demonstr ation learning framework that enables a single person to teach multiple robots. Building upon this framework\, we formalize three approaches to teaching emergent collaborative behavior based on different information sharing strategies. We provide detailed evaluations of all algorithms in multiple simulated and robotic domains\, and present a case study analy sis of the scalability of the presented techniques using up to seven rob ots. UID:DDEADA6D-5C47-424C-9EA4-1BC747970F87 TRANSP:OPAQUE URL;VALUE=URI:message:%3C49A443C7.6090502@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090305T100000 DTSTAMP:20090224T191639Z SUMMARY:Thesis Oral - Confidence-Based Robot Policy Learning from Demons tration\, Sonia Hays Chernova CREATED:20090224T191639Z DTEND;TZID=US/Eastern:20090305T110000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:5334AD82-8A09-44DB-BE32-24524F4CFAB7 DTSTART;TZID=US/Eastern:20090402T150000 URL;VALUE=URI:message:%3C14660_1237920935_n2OItT91007877_49C92C33.802070 9@cs.cmu.edu%3E DTSTAMP:20090324T190225Z SUMMARY:Gasching/Oakley SCS Distinguished Lecture: The Algorithmic Age - Bernard Chazelle CREATED:20090324T190225Z DTEND;TZID=US/Eastern:20090402T160000 LOCATION:Wean 7500 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:A brief review of the history of data warehousing and busine ss intelligence\, in the world at large and at Facebook specifically. So me highly speculative discussion of how things might evolve over the nex t few years\, as well as a specific discussion of some system design cha llenges that Cloudera is currently addressing. UID:098AC267-40D0-4914-A824-A93A89B524E9 TRANSP:OPAQUE URL;VALUE=URI:message:%3C31D3E28392D64CADBB55A1F5B5FF592F@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20090409T120000 DTSTAMP:20090409T152532Z SUMMARY:SDI / LCS Seminar -Global Information Platforms: Evolving the Da ta Warehouse\, Jeff Hammerbacher - Cloudera CREATED:20090409T152532Z DTEND;TZID=US/Eastern:20090409T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Clarke\, Emerson and Sifakis were recently awarded the Turin g Prize for "their role in developing Model-Checking into a highly effec tive verification technology\, widely adopted in the hardware and softwa re industries". Recently\, there has been considerable interest in the u se of Model Checking techniques for the automatic analysis of stochastic models arising in Systems Biology. Unfortunately\, the state space of s tochastic biological models is often too large for classical Model Check ing techniques. For these models\, a statistical approach to Model Check ing is an effective alternative. We present the first algorithm for perf orming statistical Model Checking using Bayesian Sequential Hypothesis T esting. We show that our Bayesian approach outperforms current statistic al Model Checking techniques\, which rely on tests from Classical (aka F requentist) statistics\, by requiring fewer system simulations. Another advantage of our approach is the ability to incorporate prior Biological knowledge about the model being verified. We demonstrate our algorithm on models from the Systems Biology literature and show that it enables f aster verification than state-of-the-art techniques\, even when no prior knowledge is available. We also sketch several interesting directions f or future work. UID:4CA3EA9F-5DD2-4211-BABB-9568564A63DC TRANSP:OPAQUE URL;VALUE=URI:message:%3C49E35440.90704@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090420T100000 DTSTAMP:20090413T150626Z SUMMARY:Thesis Proposal - Statistical Model Checking for Complex Stochas tic Models in Systems Biology\, Sumit Kumar Jha CREATED:20090413T150626Z DTEND;TZID=US/Eastern:20090420T110000 LOCATION:7220 Wean Hall END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:56868A5A-1D69-49EF-AD26-0E143AE9D498 DTSTART;TZID=US/Eastern:20090417T120000 DTSTAMP:20090417T143845Z SUMMARY:15th Annual MoBOT Races CREATED:20090417T143804Z DTEND;TZID=US/Eastern:20090417T130000 LOCATION:Race Course Wean Hall END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:C8C13DBD-7A46-466F-AC28-8298A9A11190 DTSTART;TZID=US/Eastern:20090417T140000 DTSTAMP:20090417T143953Z SUMMARY:Thesis Oral - Trust Me: Design Patterns for Constructing \nTrust worthy Trust Indicators\, Serge Egelman CREATED:20090417T143902Z DTEND;TZID=US/Eastern:20090417T160000 LOCATION:2101 CIC END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:44BDA4BF-5D2C-4BF8-B7D6-80F2BCB4DACA DTSTART;TZID=US/Eastern:20090424T093000 URL;VALUE=URI:message:%3C49E89356.1090106@cs.cmu.edu%3E DTSTAMP:20090417T144520Z SUMMARY:Thesis Oral - Making APIs More Usable with Improved API Designs\ , Documentation and Tools\, Jeffrey Stylos CREATED:20090417T144520Z DTEND;TZID=US/Eastern:20090424T110000 LOCATION:Wean 4625 END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:165AA902-4445-4D67-834E-0843968FC9A3 DTSTART;TZID=US/Eastern:20090417T120000 DTSTAMP:20090417T145111Z SUMMARY:POP Seminar - Paul-André Melliès CREATED:20090417T144732Z DTEND;TZID=US/Eastern:20090417T130000 LOCATION:Wean 4615A END:VEVENT BEGIN:VEVENT SEQUENCE:7 TRANSP:OPAQUE UID:F882AA57-CF0E-4284-8E7C-3933D8085350 DTSTART;TZID=US/Eastern:20090417T153000 DTSTAMP:20090417T145202Z SUMMARY:Thesis Oral - The Logical Basis of Evaluation Order and Pattern- Matching\, Noam Zeilberger CREATED:20090417T145118Z DTEND;TZID=US/Eastern:20090417T170000 LOCATION:Wean 4623 END:VEVENT BEGIN:VEVENT SEQUENCE:9 TRANSP:OPAQUE UID:D92AA1E0-BB25-4AA9-83C3-F84349A03DF0 DTSTART;TZID=US/Eastern:20090419T150000 DTSTAMP:20090417T145359Z SUMMARY:Carnegie Mellon Jazz Ensemble Concert - Jazz in the Digital Age CREATED:20090417T145257Z DTEND;TZID=US/Eastern:20090419T170000 LOCATION:Homestead Carnegie Music Hall END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:BD16F96D-C1B1-4AEE-9089-66E5ABA32D26 DTSTART;TZID=US/Eastern:20090420T163000 DTSTAMP:20090417T145508Z SUMMARY:Joint CMU-PITT Physics Colloquium — The Next Spectroscopy: New E lementary Particles\nat the Large Hadron Collider\,\nMichael Peskin CREATED:20090417T145418Z DTEND;TZID=US/Eastern:20090420T173000 LOCATION:Wean 7500 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:AA5A9F46-C108-4A21-A395-51E3EEDFC25C DTSTART;TZID=US/Eastern:20090421T143000 DTSTAMP:20090417T145603Z SUMMARY:Computational Thinking Seminar — Irfan Essa CREATED:20090417T145532Z DTEND;TZID=US/Eastern:20090421T153000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:6 DESCRIPTION:Arthur Holly Compton Distinguished Service Professor\nof Ast ronomy & Astrophysics\nChair of the Department\, University of Chicago UID:04FEE7E9-2083-4387-AAE6-075AC987E315 TRANSP:OPAQUE URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5150.html DTSTART;TZID=US/Eastern:20090421T110000 DTSTAMP:20090417T145723Z SUMMARY:2009 Buhl Lecture — Mysteries of the Dark Universe\, Edward W. R ocky Kolb CREATED:20090417T145616Z DTEND;TZID=US/Eastern:20090421T120000 LOCATION:Auditorium Mellon Institute END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:DCFF446D-77E9-4FDE-A807-AD7F5B5240E5 DTSTART;TZID=US/Eastern:20090422T163000 DTSTAMP:20090417T145810Z SUMMARY:President's Town Hall Meeting CREATED:20090417T145739Z DTEND;TZID=US/Eastern:20090422T173000 LOCATION:McConomy Auditorium UC END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Andrew Moore\, CERT senior Technical Staff Member\, will be our guest speaker at the CyLab Seminar today at 12pm. Details on the ta lk are available at http://www.cylab.cmu.edu/news_events/events/e_moore5 _11_09.html. UID:DA21521F-16D3-4778-8B28-F2BE55E8BFEF TRANSP:OPAQUE URL;VALUE=URI:message:%3C7D343235E1F14E99AB0F2B3D76C680FF@andrew.ad.cmu. edu%3E DTSTART;TZID=US/Eastern:20090511T120000 DTSTAMP:20090511T134147Z SUMMARY:CyLab Seminar: Andrew Moore "Inside Theft of Intellectual Proper ty in Organizations\, A Preliminary Model" CREATED:20090511T134147Z DTEND;TZID=US/Eastern:20090511T130000 LOCATION:INI Distributed Education Center (DEC) on the Lobby level of t he CIC building END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:7D16F37F-2D60-4C33-B03A-FBE2BB487549 DTSTART;TZID=US/Eastern:20090512T110000 URL;VALUE=URI:message:%3C4A083D9F.6010902@cs.cmu.edu%3E DTSTAMP:20090511T151852Z SUMMARY:ANDREAS TERZIS Talk (CS Professor\, Johns Hopkins University) at Intel Research Pittsburgh CREATED:20090511T151852Z DTEND;TZID=US/Eastern:20090512T130000 LOCATION:CIC Panther Hollow END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Sven Stork\, my student in the CMU/Portugal Ph.D. program\, and his co-advisor Paulo Marques will be visiting later this week and pa rt of next week. He'll give a talk on his research Monday at 3:00 in We an 8220 (the LF meeting time and room\, thanks Frank!)\n\nSven is workin g on a new approach to parallel languages:\n\n* Like functional programm ing\, the programming model naturally supports parallel execution\, up t o explicit data dependencies\n\n* Mutable state is supported\; dependenc ies between operations on that state are made explicit through the flow of permissions (which essentially generalize monads)\n\nSven (and Paulo and I) would appreciate your feedback if the topic is of interest. UID:9D60E487-46D9-409E-A073-6962ADB44607 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4A019FB8.6080505@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090511T150000 DTSTAMP:20090511T173304Z SUMMARY:[pop-group] PL/concurrency student talk CREATED:20090511T173304Z DTEND;TZID=US/Eastern:20090511T160000 LOCATION:Wean 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nAppel and McAllester devised the "step-indexed model" in order to\nexpress "semantic" proofs of type safety for use in foundational\nproof-carrying code. The basic idea is to characterize ty pe\ninhabitation as a predicate indexed by the number of steps of\ncompu tation left before "the clock" runs out. In subsequent work\,\nAhmed an d coworkers have shown how to generalize the step-indexed\nmodel to bina ry "step-indexed logical relations"\, with which one may\nreason inequat ionally about programs in languages with semantically\ncomplex type stru cture (e.g. universal\, existential\, recursive and\nmutable reference t ypes). However\, a continual annoyance in working\nwith step-indexed lo gical relations\, as well as a stumbling block to\ntheir general accepta nce\, is the tedious\, error-prone\, and\nproof-obscuring step-index ari thmetic that seems superficially to be\nan essential element of the meth od.\n\nIn this work\, we show how to reason about binary step-indexed lo gical\nrelations in a more abstract\, high-level way. Specifically\, we define\na logic LSLR\, which is inspired by Plotkin and Abadi's logic f or\nrelational parametricity\, but also supports recursively defined\nre lations by means of the modal "later" operator from Appel et al.'s\n"ver y modal model" paper. We encode in LSLR a logical relation for\nreasoni ng (in-)equationally about programs in call-by-value System F\nextended with recursive types. Using this logical relation\, we derive\na useful set of proof rules that allow us to harness the power of\nstep-indexed logical relations without counting steps.\n\nThis is joint work with Ama l Ahmed and Lars Birkedal.\n\nHost: Robert Harper\nAppointments: Christi ne Simony \n UID:13FAB5EE-6723-46F2-A04D-129BC2A91279 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4A0AE334.7060701@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090522T153000 DTSTAMP:20090513T151718Z SUMMARY:POP Seminar -Logical Step-Indexed Logical Relations\, Derek Drey er\, MPI-SWS CREATED:20090513T151718Z DTEND;TZID=US/Eastern:20090522T170000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nWe argue that passive UHF RFID technology can be combined with low-end sensing and computation platforms to realize senso r network applications that are truly ubiquitous and embedded in the phy sical world. This is because RFID brings the advantages of small\, inexp ensive and long-lived nodes to wireless sensor networks. To work towards this vision\, we introduce the notion of RFID sensor networks\, which c onsist of RFID readers and RFID-based sensor nodes that we call WISPs. I n this talk\, I will describe sample applications for RFID sensor networ ks and the Intel WISP that have been developed over the past few years. The WISP has multiple sensors and an ultra-low power microcontroller\, a nd harvests its operating power from and communicates sensor data to a s tandard (EPC Class 1 Gen 2) UHF RFID reader. I will highlight the resear ch challenges in realizing RFID sensor network applications\, such as ru nning with intermittent power and RFID protocols that are suited to sens or queries.\n \nThis is joint work with Joshua Smith\, Alanson Sample\, Dan Yeager\, Michael Buettner\, Ben Greenstein\, Polly Powledge and Rich a Prasad of Intel Research and the University of Washington. UID:3DE28670-1BA3-41B5-94D2-24191BE9433C TRANSP:OPAQUE URL;VALUE=URI:message:%3C00b001c9e395$6ea3f180$4bebd480$@cmu.edu%3E DTSTART;TZID=US/Eastern:20090604T120000 DTSTAMP:20090602T154228Z SUMMARY:SDI / LCS Seminar - Realizing RFID Sensor Networks with the Inte l WISP David Wetherall\, Intel Research and Univ. of Washington CREATED:20090602T154228Z DTEND;TZID=US/Eastern:20090604T130000 LOCATION:Wean Hall 8220 END:VEVENT BEGIN:VEVENT SEQUENCE:3 TRANSP:TRANSPARENT UID:E5456091-D831-4D1E-9CFE-951060EBFCE2 DTSTART;VALUE=DATE:20090824 DTSTAMP:20090819T163634Z SUMMARY:First Day of Classes CREATED:20090819T163624Z DTEND;VALUE=DATE:20090825 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:TRANSPARENT UID:790B636D-9467-45E5-A239-8B8B3EBDB697 DTSTART;VALUE=DATE:20090824 DTSTAMP:20090819T163703Z SUMMARY:CSD Immigration 2009 CREATED:20090819T163637Z DTEND;VALUE=DATE:20090905 END:VEVENT BEGIN:VEVENT SEQUENCE:3 TRANSP:OPAQUE UID:3A470598-67E3-41EC-980A-3E2178660F1F DTSTART;TZID=US/Eastern:20090825T130000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5388.html DTSTAMP:20090819T163802Z SUMMARY:Google Pittsburgh Seminar - Google Green Data Center Technology Best Practices and Metrics\, Todd Underwood\, Site Reliability Manager\, Google\, Inc. CREATED:20090819T163705Z DTEND;TZID=US/Eastern:20090825T140000 END:VEVENT BEGIN:VEVENT SEQUENCE:3 TRANSP:TRANSPARENT UID:8762C0D9-2859-45B7-AEFA-FC706BE4A236 DTSTART;VALUE=DATE:20090922 DTSTAMP:20090819T163859Z SUMMARY:Gates Hillman Center Dedication CREATED:20090819T163843Z DTEND;VALUE=DATE:20090923 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:TRANSPARENT UID:CCAD35F4-CA43-4840-9D86-897E7DCD2788 DTSTART;VALUE=DATE:20090929 DTSTAMP:20090819T163929Z SUMMARY:Yahoo! Hack U 2009 CREATED:20090819T163905Z DTEND;VALUE=DATE:20091004 END:VEVENT BEGIN:VEVENT SEQUENCE:5 DESCRIPTION:The meeting will focus on campus issues related to the upcom ing G20 Summit. UID:8BC02188-B702-4611-96EE-A16FBB7BA71D TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20090902T163000 DTSTAMP:20090827T153849Z SUMMARY:Open Town Meeting CREATED:20090827T153814Z DTEND;TZID=US/Eastern:20090902T173000 LOCATION:Danforth Lounge UC END:VEVENT BEGIN:VEVENT SEQUENCE:8 TRANSP:OPAQUE UID:2564EF6F-A3E9-45D4-9CB4-CE23582EDBA3 DTSTART;TZID=US/Eastern:20090910T160000 URL;VALUE=URI:http://calendar.cs.cmu.edu/dlseries/5403.html DTSTAMP:20090827T180550Z SUMMARY:SCS Distinguished Lecture Series\nTHE COMPUTERS AND THOUGHT AWAR D LECTURE 2009: How Optimized Environmental Sensing Helps Address Inform ation Overload on the Web\, Carlos Guestrin CREATED:20090827T153905Z DTEND;TZID=US/Eastern:20090910T170000 LOCATION:Rashid Auditorium (GHC 4401) END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:http://calendar.cs.cmu.edu/dlseries/5403.html UID:E890D76D-4A35-49C0-80D9-8B8C65586A8B TRANSP:OPAQUE URL;VALUE=URI:message:%3CC6BC0BA6.C835%25abbyross@andrew.cmu.edu%3E DTSTART;TZID=US/Eastern:20090911T120000 DTSTAMP:20090827T154133Z SUMMARY:Andy Awards CREATED:20090827T154133Z DTEND;TZID=US/Eastern:20090911T230000 LOCATION:McConomy Auditorium\, UC END:VEVENT BEGIN:VEVENT SEQUENCE:9 TRANSP:OPAQUE UID:57987A67-C7C9-4937-BCCB-AA4A615357FD DTSTART;TZID=US/Eastern:20090902T160000 DTSTAMP:20090831T155334Z SUMMARY:CSD Faculty Meeting EXDATE;TZID=US/Eastern:20100106T160000 EXDATE;TZID=US/Eastern:20091104T160000 CREATED:20090831T155130Z DTEND;TZID=US/Eastern:20090902T170000 LOCATION:GHC 6115 RRULE:FREQ=MONTHLY;INTERVAL=1;UNTIL=20100805T035959Z;BYDAY=1WE END:VEVENT BEGIN:VEVENT SEQUENCE:11 DESCRIPTION:Just a reminder\, there will be a CSD Faculty meeting this W ednesday\, October\n7. We will begin with light snacks at 3:40 p.m. fol lowed by the meeting at\n4:00 p.m. UID:57987A67-C7C9-4937-BCCB-AA4A615357FD TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20091021T160000 DTSTAMP:20091007T144537Z SUMMARY:CSD Faculty Meeting CREATED:20090831T155130Z DTEND;TZID=US/Eastern:20091021T170000 LOCATION:GHC 6115 RECURRENCE-ID;TZID=US/Eastern:20091007T160000 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:472F50B3-EE45-455B-8AB2-69D311FF75A6 DTSTART;TZID=US/Eastern:20090904T153000 URL;VALUE=URI:message:%3C4A9D1107.7050001@cs.cmu.edu%3E DTSTAMP:20090901T153610Z SUMMARY:Speaking Skills Talk - Certifying C Compiler with Fat pointers\, Miguel Silva CREATED:20090901T153610Z DTEND;TZID=US/Eastern:20090904T163000 LOCATION:GHC 9115 END:VEVENT BEGIN:VEVENT SEQUENCE:4 DESCRIPTION:http://seminar.living-environments.net/ \n\nOf all the renew able energy sources\, wind exhibits the greatest promise: production cos ts per kilowatt-hour are close to those of fossil energy\; potential win d power far exceeds total U.S. power demand\; wind is rapidly becoming a proven technology as illustrated by countries such as Spain\, Germany a nd Denmark deriving 10-30% of their electricity from wind farms alone. Y et\, many challenges remain: the capital cost of wind farms is large\; i ntegrating wind sources into the power grid requires substantial upgrade s\; wind power is variable and power storage and buffering is quite diff icult.\n\nThe presentation focuses on optimizing wind power production a t all levels:\n\n· Selection of equipment and vendor\n\n· Adaptation to local topography and weather patterns\n\n· Grid- level optimization (distance\, wattage\, buffering\, …)\n\n· Nat ional-level optimization (complementary\, security\, …)\n\n· NPV “lifetime” capital-cost and operations optimization\n\nThe above help t o drive down costs further so as to ensure wind practicality\, and to pr ovide sufficient and balanced electrical power where needed. Moreover\, the ability to model\, analyze and optimize wind projects including “wh at-if” analyses can provide a trusted third-party evaluator for the powe r industry\, vis-à-vis vested-interest wind turbine vendors and provider s of other power generation technologies. UID:F7F5957C-050A-4881-ACCC-0CDC70EAD09A TRANSP:OPAQUE URL;VALUE=URI:message:%3C4A9E7D6B.8020108@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090911T140000 DTSTAMP:20090903T145103Z SUMMARY:Sustainability and Computer Science Seminar — Wind Power: Optimi zation at All Levels\, Jaime Carbonell\, CMU CREATED:20090902T151954Z DTEND;TZID=US/Eastern:20090911T150000 LOCATION:Rashid Auditorium\, GHC 4401 END:VEVENT BEGIN:VEVENT SEQUENCE:11 TRANSP:OPAQUE UID:FE4CC721-7000-406D-AA81-C708BA0FFF85 DTSTART;TZID=US/Eastern:20091009T013000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5454.html DTSTAMP:20091005T171906Z SUMMARY:Sustainability Seminar Series — Customizing Commute Ecology: A c ommunity-empowered road for electric vehicles\, Illah Nourbakhsh CREATED:20090902T152101Z DTEND;TZID=US/Eastern:20091009T023000 LOCATION:GHC Rashid Auditorium (4401) END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Thesis Oral\n\nTitle: Matching Markets: Design and Analysis\ n\nAbstract:\n\nA market consists of buyers and sellers of some commodit y\, say a DVD movie. In this thesis\, we assume the role of market opera tor. Our goal is to ensure that the market has certain desirable propert ies\, such as efficiency and fairness. We achieve these properties by de signing rules for how buyers and sellers can perform transactions. Our m ain focus will be on matching markets. Some examples of matching markets include online DVD rental\, pay-per-click keyword auctions and kidney e xchanges.\n\nThesis Committee:\nR. Ravi\, Chair\nAlan Frieze\nLuis von A hn\nDavid Manlove\, University of Glasgow UID:7810350A-D081-4297-889B-28CB81094268 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AA10917.8090808@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20090911T093000 DTSTAMP:20090904T141931Z SUMMARY:Thesis Oral - Matching Markets: Design and Analysis\, David Abra ham CREATED:20090904T141931Z DTEND;TZID=US/Eastern:20090911T103000 LOCATION:GHC 8115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:63782B8D-5D0D-488E-8541-CDE130CDFDF4 DTSTART;TZID=US/Eastern:20090923T130000 URL;VALUE=URI:message:%3C003e01ca3c69$34abafe0$9e030fa0$@cmu.edu%3E DTSTAMP:20090923T161916Z SUMMARY:SPECIAL SDI / LCS Seminar held jointly with IRHPIT and CSSI (wit h roundtable discussion to follow from 2 - 3 in same room) CREATED:20090923T161916Z DTEND;TZID=US/Eastern:20090923T140000 LOCATION:GHC7101 END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:48BCA940-A8B0-4985-9430-C3E4AE8CE01C DTSTART;TZID=US/Eastern:20090930T150000 DTSTAMP:20090924T195916Z SUMMARY:CSD Reappointment & Promotion Meeting CREATED:20090924T195825Z DTEND;TZID=US/Eastern:20090930T170000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:4 DESCRIPTION:PA Governor Ed Rendell\, Allegheny County Executive Dan Onor ato\, Pittsburgh Mayor Luke Ravenstahl and Carnegie Mellon President Jar ed Cohon welcome CT &T\, a South Korean electric car manufacturer for a visit to Carnegie Mellon University to discuss potential research collab orations with prominent faculty. There will also be a brief driving demo nstration on campus.\n \nThe event will begin at 11:00 a.m. on Friday\, September 25\, 2009 at the 5th Floor entrance plaza to the Gates Buildin g (near the Warner Hall parking lot\, directly behind the Purnell Center ). Light lunch reception to follow.\n \nThese two-passenger electric cit y vehicles are specifically designed for high density urban areas. They reach speeds of up to 40 miles per hour and can be recharged by a standa rd electric wall outlet. Estimated monthly operational costs are $7.00. CT &T is seeking out potential locations in the region for manufacture a nd or manufacturing of sub components to be used in the vehicle. CT & T is the largest electric vehicle manufacturer in the world today. One 8-h our shift can produce 160\,000 vehicles per shift in their four separate factories. They are currently selling electric cars in South Korea\, Ch ina\, and Japan and are now moving to North America. UID:3E626771-3D68-4067-8B91-223A6CB11D61 TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20090925T110000 DTSTAMP:20090925T142157Z SUMMARY:CT&T Electric Car Event CREATED:20090925T142055Z DTEND;TZID=US/Eastern:20090925T120000 LOCATION:5th Floor entrance GHC END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:6F0B0A2C-A627-417C-B7E0-D40433E946C8 DTSTART;TZID=US/Eastern:20091002T153000 URL;VALUE=URI:message:%3C2fe07cb50909250745k77e8ca49l714e00e1d1958edf@ma il.gmail.com%3E DTSTAMP:20090925T161045Z SUMMARY:Making Machines Listen - Paris Smaragdis CREATED:20090925T161045Z DTEND;TZID=US/Eastern:20091002T163000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:7 TRANSP:OPAQUE UID:2B716C19-0068-462C-A5D5-A59FA1410525 DTSTART;TZID=US/Eastern:20090929T120000 DTSTAMP:20090929T140731Z SUMMARY:Computer Systems Lunch CREATED:20090929T140701Z DTEND;TZID=US/Eastern:20090929T130000 LOCATION:GHC 6121 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nIn most computer systems users' access to resou rces is restricted based on authorization policies. It well known that l ogic is an appropriate medium for representing\, understanding\, and enf orcing authorization policies\, yet despite several years of pragmatic w ork on the subject\, the foundations of relevant logics remain unexplore d and poorly understood. It is in this realm that the work of this thesi s lies\; the thesis explores the theory of logics for expressing authori zation policies as well as applications of the theory in practice. In do ing so\, it makes three foundational and technically challenging contrib utions.\n\nFirst\, the thesis introduces proof theory and metatheory in the context of authorization logics\, illustrated through a new logic BL . In particular\, structural proof-theoretic systems of natural deductio n and sequent calculus are investigated and their importance explained. Pragmatic problems like proof verification and automatic proof search ar e then addressed using the sound foundations of proof theory.\n\nSecond\ , the thesis considers a logical treatment of dynamism in authorization policies and in particular\, logical constructs for representing authori zations depending on system state\, consumable credentials\, and explici t time are presented. Further\, a practical\, efficient\, and provably c orrect mechanism for their enforcement is developed. The mechanism is ba sed on a combination of proofs and cryptographic capabilities.\n\nThird\ , the practical usefulness of the proof theory and the enforcement mecha nism is demonstrated through an implementation of the same in a file sys tem\, PCFS. It is shown through measurements that file access in PCFS is extremely efficient.\n\nIn addition\, the thesis includes a detailed ca se study that formalizes in BL the policies used to control access to cl assified information in the U.S.\, and explains how the policies may be enforced using PCFS.\n\nThesis Committee:\nFrank Pfenning\, Chair\nLujo Bauer\nAnupam Datta\nRobert Harper\nMartín Abadi\, UCSC/Microsoft Resear ch\n UID:901AD5F4-70E1-4FA1-9D93-C8B252D053E8 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AC00DDF.4050104@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091001T150000 DTSTAMP:20090929T140849Z SUMMARY:Thesis Oral - Proof Theory for Authorization Logic and its Appli cation to a Practical File System\, Deepak Garg CREATED:20090929T140849Z DTEND;TZID=US/Eastern:20091001T160000 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Lecturer on Computer Science\nhttp://www.cs.harvard.edu/~mal an/aboutme.shtml\n\n*ABSTRACT*\n\n"Computer Science 50 is Harvard Colleg e’s introductory course for majors and\nnon-majors alike\,a one-semester amalgam of courses generally known as CS\n1 and CS 2 taught mostly in C. CS 50 is required of majors\, but most of\nthe course’s students are non-majors. The course has never lacked for good\nteachers or content\, but we dare say it has suffered some problems of\nperception and\, like most schools\, declining enrollment.\n\nIn Fall 2007\, we set out to co mbat both. Not only did we "modernize" the\ncourse's assignments\, brin ging the focus of each more in line with domains\nmore familiar to stude nts\, we also took care to make the course more\naccessible to those sel f-described as "less comfortable" with technology.\nBut we did not have to sacrifice the course's fundamentals or rigor\; the\ncourse has preser ved its reputation as having one of the highest workloads\nat Harvard.\n \nIn Fall 2006\, enrollment was 132. In Fall 2007\, it was 282 and\, in Fall\n2008\, 330. I present in this talk *The New* *CS 50.*"\n\n*BIO*\n \nDavid Malan currently hold an appointment as Lecturer on Computer Scie nce\nat Harvard College for the School o f\nEngineering and Applied Sciences. He\n received his A.B.\, S.M.\, and Ph.D. in Computer Science from the\nsame in 1999\, 2004\, and 2007\, respectively. He teaches the College's\nintr oductory computer science course\, Computer Science 50\, otherwise known \nas CS 50 . He also serve as Chief Information Office r\nfor Mindset Media and teach at Harvard Extension School and\nHarvard Summer School. His research in\ngraduate school focused primarily on cybersecurity and computer forensics.\nWhile in graduate school\, he founded two startups Crimson Tutors and\nDiskas ter. He worked part-time for the Middlesex District Attorney's\nOffice\n as a forensic investigator. And\, on the side\, he volunteered as an EMT -B\nfor MIT-EMS. He continues to volunteer as an EMT-B for the American Red\nCross.\n=========================================================== ==\n\nIf you would like to meet with David while he is here\, please con tact Barb\nGrandillo at bag@cs.cmu.edu and she will be glad to help you get you on\nhis schedule. There are few appointments still available on Tuesday.\n\nFor more information on the colloquium on Computer Science P edagogy\,\nplease visit\nhttp://www.csd.cs.cmu.edu/colloquium/\n UID:29DF3FB9-ACF4-4A5C-9E55-F6FFECF08C2E TRANSP:OPAQUE URL;VALUE=URI:message:%3Cd91548392886a06d6dd6493cbb670815.squirrel@webma il.andrew.cmu.edu%3E DTSTART;TZID=US/Eastern:20091006T150000 DTSTAMP:20090929T172713Z SUMMARY:Colloquium on Computer Science Pedagogy - The New CS 50\, David Malan\, Harvard College CREATED:20090929T172713Z DTEND;TZID=US/Eastern:20091006T160000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:4 DESCRIPTION:e 3rd Annual Ice Cream Social is this coming Monday\, Octobe r 5\, from Noon to 1 p.m. in the University Center’s Connan Room. We ar e once again serving Dave & Andy’s ice cream (vanilla\, chocolate and pu mpkin pie) as well as strawberry champagne sorbet. There will also be a raffle for anyone who signs up for our email list – prizes include pass es for cultural attractions\, gift cards and Eat ‘n Park Smiley cookies. UID:5BE1AFC2-7657-43F8-9532-3107CC022A3B TRANSP:OPAQUE DTSTART;TZID=US/Eastern:20091005T120000 DTSTAMP:20091005T150744Z SUMMARY:Staff Council Ice Cream Social CREATED:20091005T150708Z DTEND;TZID=US/Eastern:20091005T130000 LOCATION:UC Connan Room END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:2E95913F-35D1-42E9-9464-316C680F711D DTSTART;TZID=US/Eastern:20091005T120000 URL;VALUE=URI:message:%3C005101ca45c3$ec7cedb0$c576c910$@cmu.edu%3E DTSTAMP:20091005T150846Z SUMMARY:CyLab Seminar CyLab seminar today: Armand Makowski-"Recent Resul ts for Random Key Graphs:Connectivity\, Triangles\, etc. " CREATED:20091005T150846Z DTEND;TZID=US/Eastern:20091005T130000 LOCATION:DEC CIC END:VEVENT BEGIN:VEVENT SEQUENCE:3 DESCRIPTION:CS Undergraduate (15-1xx - 15-6xx) - Angie Brookins (angieb@ cs)\nCS Graduate (15-7xx - 15-8xx) Debbie Cavlovich (deb@cs)\nMSE (17-6x x) - Ellen Saxon (ens@cs)\nLTI (11-6xx - 11-7xx) - Stacey Young (staceyy +@cs.cmu.edu )\nRobotics (16-1xx - 16-8xx) Suzanne Muth (lyonsmuth@cmu.e du)\nMLD(10-xxx) - Diane Stidle (diane@cs)\nISRI SE-PhD (17-7xx\, 17-8xx and 17-2xx) - Connie Herold\nISRI MSIT-EBusiness (96-xxx) Amber Vivis ( albrown@andrew.cmu.edu)\n\nPlease send mail to the appropriate administr ator ASAP with the\nfollowing information in order for them to submit yo ur textbook order:\n\ntitle:\nauthor:\npublisher:\nISBN:\nedition:\ncour se number/section:\nsecond instructors name if you are co-teaching:\nnum ber of books to order:\nrequired or optional text?:\nnumber of desk copi es you will need: UID:5406668A-9978-49C3-AF7B-4078A471D708 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AC9F786.2030908@cs.cmu.edu%3E DTSTART;VALUE=DATE:20091106 DTSTAMP:20091005T151028Z SUMMARY:Reminder: Textbook Orders - Spring 2010 and Mini 4 DEADLINE CREATED:20091005T150954Z DTEND;VALUE=DATE:20091107 LOCATION:Univ. Bookstore END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:http://www.pdl.cmu.edu/SDI/2009/100609.html\n \nSPEAKER:\nEr ic Baldeschwieler\nVP Software Engineering\, Hadoop Yahoo!\n \nTITLE:\nH adoop at Yahoo - learnings and challenges\n \nABSTRACT:\n \nYahoo is the primary developer of Hadoop and the runs the largest Hadoop clusters in the world. I will cover the scale of Hadoop usage at Yahoo and discuss recent challenges we have faced. I will end with a list of open research opportunities in Hadoop.\n \nBIO:\n \nEric Baldeschwieler has been a me mber of Yahoo!s web-search/cloud teams since 1996 and the founder of the Yahoo! team that has taken Hadoop from a 20 node prototype to a 25\,000 + node service that powers key production applications across Yahoo! Pre viously he has worked on video Games\, video special effects systems\, a nd 3D rendering products. He holds a BS in Applied Math (CS) from Carneg ie Mellon and an MS in CS from UC Berkeley.\n \nSDI / LCS Seminar Questi ons?\nKaren Lindenfelser\, 86716\, or visit www.pdl.cmu.edu/SDI/ UID:1AD68457-279D-459B-BA9E-27438F32E862 TRANSP:OPAQUE URL;VALUE=URI:message:%3C8C88C867770D42F6BE151FB9D19BFF8B@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20091006T120000 DTSTAMP:20091005T151153Z SUMMARY:SPECIAL SDI / LCS Seminar - Eric Baldeschwieler - Yahoo! CREATED:20091005T151153Z DTEND;TZID=US/Eastern:20091006T130000 LOCATION:CIC 2101 END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:A86128A1-DAB9-4460-A1B6-AEC4426AE9A3 DTSTART;TZID=US/Eastern:20091006T153000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5478.html DTSTAMP:20091005T171211Z SUMMARY:Intelligence Seminar - AI After Dark: Computers Playing Poker\, Michael Bowling\, University of Alberta CREATED:20091005T171054Z DTEND;TZID=US/Eastern:20091006T163000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:FFADE236-7AA0-49AD-A86F-4FBC74A9DA6F DTSTART;TZID=US/Eastern:20091007T160000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5479.html DTSTAMP:20091005T171352Z SUMMARY:HCII Seminar \nEdward J. Coyle\, Arbus Chair for the Integration of Research and Education\nGeorgia Institute of Technology - Bridging t he Intellectual Divide: \nIntegrating Research and Education via the\nVe rtically-Integrated Projects Program CREATED:20091005T171255Z DTEND;TZID=US/Eastern:20091007T170000 LOCATION:NSH Mauldin Auditorium END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:50A9AA75-4626-4E78-9FC8-E40394ED5F66 DTSTART;TZID=US/Eastern:20091008T140000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5475.html DTSTAMP:20091005T171530Z SUMMARY:Data Analysis Project Presentation - Spectral Kernel Methods for Protein Fold Recognition\, James Sharpnack CREATED:20091005T171408Z DTEND;TZID=US/Eastern:20091008T150000 LOCATION:Steinberg Auditorium (A53) Baker Hall END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:83668B98-C77C-4600-B23F-7CB6D207E9A5 DTSTART;TZID=US/Eastern:20091008T160000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5483.html DTSTAMP:20091005T171729Z SUMMARY:Robotics Seminar - Robot Artsists-Robot Dancer and Robot Painter \, Sunsuke Kudoh\, Institute of Industrial Science\, Univ. of Tokyo CREATED:20091005T171546Z DTEND;TZID=US/Eastern:20091008T170000 LOCATION:GHC Reddy Conference Room (4405) END:VEVENT BEGIN:VEVENT SEQUENCE:7 TRANSP:OPAQUE UID:5EF68C06-0DD1-4B42-A711-ED2C94E02895 DTSTART;TZID=US/Eastern:20091009T110000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5485.html DTSTAMP:20091005T172122Z SUMMARY:Joint CMU-Pitt Ph.D. Program in Computational Biology Seminar - Synthesizing and Simplifying Biological Networks from Pathway Level Info rmation\, Bhaskar DasGupta\, Computer Science Dept.\, Univ. of Illinois at Chicago CREATED:20091005T171924Z DTEND;TZID=US/Eastern:20091009T120000 LOCATION:GHC 8102 END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:A77F8680-5066-401F-BB03-9E5BAF050981 DTSTART;TZID=US/Eastern:20091013T120000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5480.html DTSTAMP:20091005T172257Z SUMMARY:Speaking Skills - BBM: Bayesian Browsing Model from Petabyte-sca le Data\, Fan Guo CREATED:20091005T172203Z DTEND;TZID=US/Eastern:20091013T130000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:E4D239EB-06DD-4AAA-A36B-584F9A434698 DTSTART;TZID=US/Eastern:20091015T183000 DTSTAMP:20091005T172340Z SUMMARY:CSD Login Ball CREATED:20091005T172313Z DTEND;TZID=US/Eastern:20091015T213000 END:VEVENT BEGIN:VEVENT SEQUENCE:3 TRANSP:TRANSPARENT UID:9B241CB8-D5E1-433E-9502-26FA742CC686 DTSTART;VALUE=DATE:20091016 DTSTAMP:20091005T172419Z SUMMARY:Mid-semester Break — No Classes CREATED:20091005T172357Z DTEND;VALUE=DATE:20091017 END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:737FA83F-DC77-45CD-846A-BC8088FDD905 DTSTART;TZID=US/Eastern:20091029T170000 URL;VALUE=URI:http://www.techbridgeworld.org/Interactive2009/ DTSTAMP:20091005T172519Z SUMMARY:TechBridge World 3rd Interactive Event CREATED:20091005T172434Z DTEND;TZID=US/Eastern:20091029T190000 LOCATION:NSH Perlis Atrium END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:88DAA308-2146-4BEF-BCFC-DAB601ADE755 DTSTART;TZID=US/Eastern:20091029T163000 DTSTAMP:20091005T172638Z SUMMARY:Alumni Distinguished Achievement Award Recipient Lecture - Willi am (Red) L. Whittaker CREATED:20091005T172549Z DTEND;TZID=US/Eastern:20091029T173000 LOCATION:GHC Rashid Auditorium END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n Every new OS needs to support legacy softw are. Virtualization is one technique to solve this problem. Unfortunate ly virtualization can have a negative impact on the security of an appl ication\, as it adds a new layer of indirection to the TCB.\n \n I n this talk I will present Vancouver on Nova\, a VMM on a thin hyperviso r\, that allows to run unmodified OSes. Due to careful design and the n ow widely available hardware support for virtualization we can achieve a TCB which is approximately an order of magnitude smaller than comparabl e solutions. UID:D459A5FE-FCD2-40D9-993D-B7915215DD49 TRANSP:OPAQUE URL;VALUE=URI:message:%3C00ae01ca45e1$0ea08740$2be195c0$@cmu.edu%3E DTSTART;TZID=US/Eastern:20091007T120000 DTSTAMP:20091005T174628Z SUMMARY:CyLab Seminar Seminar - Berhhard Kauer "Vancouver on Nova - Virt ualization with a small TCB" CREATED:20091005T174628Z DTEND;TZID=US/Eastern:20091007T130000 LOCATION:INI DEC CIC END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:05CA3969-54DE-4ABC-9843-124710D946F1 DTSTART;TZID=US/Eastern:20091009T141500 URL;VALUE=URI:message:%3C13527_1254904428_n978Xj7w026479_Pine.LNX.4.63.0 910060951360.22523@islay.ius.cs.cmu.edu%3E DTSTAMP:20091007T134523Z SUMMARY:Photorealistic Rendering: Crossing the Uncanny Valley— Academy a ward winner Henrik Wann Jensen CREATED:20091007T134523Z DTEND;TZID=US/Eastern:20091009T153000 LOCATION:NSH 3305 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:7EF6CAED-EB21-43EA-8804-B1E53CC06408 DTSTART;TZID=US/Eastern:20091015T163000 DTSTAMP:20091007T144910Z SUMMARY:ECE Distinguished Lecture - Yale Patt\, UT Austin CREATED:20091007T144723Z DTEND;TZID=US/Eastern:20091015T173000 LOCATION:Scaife Hall Auditorium END:VEVENT BEGIN:VEVENT SEQUENCE:5 DESCRIPTION:PDF of abstract: http://www.ece.cmu.edu/news/seminar/2009/fa ll/hwu_10_08_09.pdf UID:93B08027-3DF5-4E46-924B-90F4AFD3D150 TRANSP:OPAQUE URL;VALUE=URI:http://www.ece.cmu.edu/news/seminar/%23sem-10-15 DTSTART;TZID=US/Eastern:20091008T110000 DTSTAMP:20091007T145050Z SUMMARY:ECE Lecture - Many-core Parallel Computing Research at the Illin ois UPCRC\, Wen-mei Hwu\, Urbana-Champaign CREATED:20091007T144936Z DTEND;TZID=US/Eastern:20091008T120000 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:F61488E0-AAEA-47C8-8A28-DB4254EFD4B0 DTSTART;TZID=US/Eastern:20091009T133000 URL;VALUE=URI:message:%3C005e01ca48f8$9daa8d90$d8ffa8b0$@cmu.edu%3E DTSTAMP:20091009T160402Z SUMMARY:CyLab Seminar - Dr. Nalini Ratha "Performance evaluation of Regi stration based and Registration-free revocable biometrics templates" CREATED:20091009T160402Z DTEND;TZID=US/Eastern:20091009T143000 LOCATION:INI Distributed Education Center (DEC) on the Lobby level of th e CIC building END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:C202A7AB-A57D-46B8-855F-9923308FEFA3 DTSTART;TZID=US/Eastern:20091013T120000 URL;VALUE=URI:message:%3C4AD47606.4080902@cs.cmu.edu%3E DTSTAMP:20091013T135458Z SUMMARY:Computer Systems Lunch CREATED:20091013T135458Z DTEND;TZID=US/Eastern:20091013T130000 LOCATION:GHC 6121 END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:Abstract:\n\nAlthough C is a fairly old language\, today it still remains as one of the most used languages with a significant amoun t of code written that is being utilized by modern software and modern l anguages. Unfortunately\, C is an unsafe language that gives users a gre at deal of flexibility in terms of memory management and does little to enforce the correct use of said memory. Furthermore\, the continuous inc rease in code size and complexity and the growing number of untrusted so urces make it essential to be able to ensure code safety. Our goal is to create a C compiler that adds runtime checks to the code to enforce mem ory safety and certifies that its output is type safe. This compiler sho uld generate code that runs without major overhead and without raising e xceptions whenever its input is legal.\n\nIn partial fulfillment of the CSD speaking skills requirement. UID:419859C6-9C79-4ECB-850D-4A88A01357F1 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AD473EF.6080900@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091027T160000 DTSTAMP:20091013T140418Z SUMMARY:Speaking Skills Talk - Certifying compilation for C\, Miguel Sil va CREATED:20091013T140353Z DTEND;TZID=US/Eastern:20091027T170000 LOCATION:GHC 8102 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nPredictive approaches for fMRI data analysis th at integrate fMRI data from multiple subjects and multiple studies are d esirable because they can potentially leverage more data to make better predictions\, and in addition\, inform us about similarities and differe nces that exist across different human subjects and different studies. H owever\, in order to be successful\, these approaches must deal with var iations that exist in different brains both anatomically and functionall y. In this talk\, I present an approach to integrate multiple fMRI datas ets in the context of predictive fMRI data analysis. The approach utiliz es canonical correlation analysis (CCA) to find common dimensions among the different datasets\, implicitly taking into account the anatomic and functional variations present in the data. We apply the approach to the task of predicting brain activations for unseen concrete-noun words usi ng multiple-subject datasets from two related fMRI studies. The proposed approach yields better prediction accuracies than those of an approach where each subject's data is analyzed separately.\n\nPresented in Partia l Fulfillment of the CSD Speaking Skills Requirement.\n UID:6489AD32-A5E2-46A1-BF5B-2FB31FB7A817 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AD47341.2010102@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091023T120000 DTSTAMP:20091013T140512Z SUMMARY:Speaking Skills Talk - Integrating Multiple-Subject Multiple-Stu dy fMRI Datasets Using Canonical Correlation Analysis\, Indrayana Rustan di CREATED:20091013T140512Z DTEND;TZID=US/Eastern:20091023T130000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:IVAN SUTHERLAND\nVisiting Scientist\nPortland State Universi ty\nand founder\, "Asynchronous Research Center" (ARC)\n\n\nFleet\, Infi nity & Marina\nAdam Megacz (Berkeley) and Ivan Sutherland (Portland Stat e)\n\nThis talk describes a radically different architecture for computi ng called Fleet. \nFleet accepts the limitations to computing imposed by physics: moving data costs more energy\, \nmore delay\, and more chip a rea than the arithmetic and logical operations ordinarily called "comput ing." Fleet puts the programmer firmly in charge of the most costly reso urce: communication. Fleet treats arithmetic and logical operations as s ide effects of where the programmer sends data.\n\nFleet achieves high p erformance through fine grain concurrency. Everything Fleet does is conc urrent \nat the lowest level\; programmers who wish sequential behavior must program it explicitly. Fleet presents a stark contrast to today's m ulti-core machines in which programmers seek concurrency in an inherentl y sequential environment.\n\nThe Fleet architecture uses a uniform switc h fabric to simplify chip design. A few thousand identical \ncopies of a configurable interface will connect a thousand or so repetitions of bas ic arithmetic\,logical\, input-output\, and storage units to the switch fabric. The uniform switch fabric and the identical configurable interfa ces will simplify many of the hard parts of designing the computing elem ents themselves.\n\nBoth software and FPGA simulators of a Fleet system are available at UC Berkeley. Berkeley students have written a variety o f Fleet programs\; their work helped to define what the configurable int erface between computing and communication must do. A simple compiler co nfigures both source and destination to provide flow-controlled communic ation. We expect work on a higher-level language for Fleet to appear soo n as a Berkeley PhD dissertation.\n\nLast year we built a 90 nanometer T SMC test chip\, called Infinity\, at Sun Microsystems. Infinity \ndemons trated the switch fabric running at about 4 GHz. We now have a new test chip\, called Marina\, \nalso in 90-nanometer TSMC sponsored by Sun. Mar ina shows correct operation of the configurable switch fabric interface. Together Infinity and Marina give us confidence to build a complete Fle et. We seek participation from sponsors\, computer scientists\, and hard ware designers.\n\n\nSPEAKER BIO\nIvan Sutherland is a Visiting Scientis t at Portland State University where he and Marly Roncken have recently established the Asynchronous Research Center (ARC). The ARC occupies bot h physical and intellectual space half way between the Computer Science (CS) and Electrical and Computer Engineering (ECE) departments at the un iversity. The ARC seeks to free designers from the tyranny of the clock by developing better tools and teaching methods for design of self-timed systems. Prior to moving to Portland\, Ivan spent 25 years as a Fellow at Sun Microsystems. A 1959 graduate of Carnegie Tech\, Ivan got his PhD at MIT in 1963 and has taught at Harvard\, The University of Utah\, and Caltech. Ivan is a member of the National Academy of Engineering and t he National Academy of Sciences. UID:AFFD1FAB-985E-4B5D-88FA-45A8D3ACE728 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AD49376.8050804@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091015T150000 DTSTAMP:20091013T152851Z SUMMARY:Special SCS Distinguished Lecture: Fleet\, Infinity & Marina\, I van Sutherland CREATED:20091013T152851Z DTEND;TZID=US/Eastern:20091015T170000 LOCATION:GHC Rashid Auditorium END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:We are happy to inform you that we have made arrangements to have Melissa Hathaway's talk webcasted into the CIC INI DEC tomorrow (W ednesday\, October 14th) at 12:00 noon.\n \nWe are very sorry that we ca nnot accommodate the students for lunch in Rangos as the space is limite d. However\, we will have pizza available in the CIC INI DEC if you are interested in attending this webcast.\n \nWhen: Wednesday\, October 14 th\nWho: Melissa Hathaway\, Keynote Speaker for CyLab Corporate Confer ence\nTime: 12:00 Noon\nWhere: CIC INI DEC UID:BEDD99D6-C2E9-4EE5-8FA7-62204A6ED3DA TRANSP:OPAQUE URL;VALUE=URI:message:%3C004501ca4c25$fb57b2d0$f2071870$@cmu.edu%3E DTSTART;TZID=US/Eastern:20091014T120000 DTSTAMP:20091013T165929Z SUMMARY:CyLab Seminar Corporate Partners Webcast CREATED:20091013T165929Z DTEND;TZID=US/Eastern:20091014T130000 LOCATION:CIC INI DEC END:VEVENT BEGIN:VEVENT SEQUENCE:6 DESCRIPTION:The CyLab Partners Conference is an annual gathering of CyLa b's corporate partners to meet with CyLab researchers and review their c urrent projects. UID:B6A28507-15A8-4DEE-8BB7-428F139AE84C TRANSP:TRANSPARENT URL;VALUE=URI:http://www.cylab.cmu.edu/news_events/events_all.html DTSTART;VALUE=DATE:20091014 DTSTAMP:20091013T172220Z SUMMARY:Cylab Parnters Conference CREATED:20091013T172055Z DTEND;VALUE=DATE:20091017 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:http://www.cylab.cmu.edu/news_events/events/e_sion10_19_09.h tml UID:E7152382-0AA3-4597-91DD-1BE9F713BF8E TRANSP:OPAQUE URL;VALUE=URI:message:%3C001301ca4dcc$f241b4e0$d6c51ea0$@cmu.edu%3E DTSTART;TZID=US/Eastern:20091019T120000 DTSTAMP:20091019T134545Z SUMMARY:CyLab Seminar — Security vs Costs & Energy in Clouds\, Radu Sion \, Stony Brook University CREATED:20091019T134545Z DTEND;TZID=US/Eastern:20091019T130000 LOCATION:INI Distributed Education Center (DEC) on the Lobby level of th e CIC building END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nCompanies that measure their infrastructure in me gawatts pay close attention to Power Usage Effectiveness\, but PUE is on ly part of the equation. Mega datacenters are often internally plagued w ith micro-infrastructures. Fragmentation may result from disparate workl oad needs (e.g.\, mapreduce vs streaming)\, physical limitations (e.g.\, cable distance)\, technology constraints (e.g.\, switch port density)\, or even policy (e.g.\, security). This talk explores some of Yahoo!’s c hoices along the path to defrag’ing the datacenter.\n \nBIO:\nAdam is he ad of infrastructure architecture at Yahoo!. Prior to Yahoo!\, Adam was the lead network architect at Inktomi Corporation. Prior to Inktomi\, Ad am worked 100+ hour weeks at a startup (like everyone else during the bu bble). Adam holds BS and MS degrees in engineering from the University o f Colorado.\n \nSDI / LCS Seminar Questions?\nKaren Lindenfelser\, 86716 \, or visit www.pdl.cmu.edu/SDI/ UID:94FC64B6-5B7E-40EC-915A-68B36DF9FC66 TRANSP:OPAQUE URL;VALUE=URI:message:%3C299AA9B9E7D84B45A6C6AB5D45B45CF5@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20091019T150000 DTSTAMP:20091019T145523Z SUMMARY:SPECIAL SDI / LCS Seminar - Defrag’ing the Datacenter\, Adam Bec htel\, Yahoo! CREATED:20091019T145523Z DTEND;TZID=US/Eastern:20091019T160000 LOCATION:CIC 2101 END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:ABSTRACT:\nPublic displays have become pervasive with the wi despread deployment of plasma screens\, LCD panels and projection system s. Shopping malls\, sports stadiums and city squares frequently feature large numbers of displays of varying size and complexity. However these displays are typically part of small isolated networks consisting of a h andful of displays under a single management domain. Our group's aim is to create a single global network of displays that is open to new applic ations and content from many sources - a Global Display Network. In effe ct\, we are looking to create the display equivalent of the Internet - a single global medium for sharing information that has revolutionized th e way we live our lives. A Global Public Display Network has the potenti al to change every public space - from environments in which information is pushed to passers-by in the form of adverts to spaces that can be ta ilored to reflect the hopes\, aspirations and interests of its occupants using content and applications created anywhere in the global network. In this talk I will focus on the infrastructure created and lessons lear ned from the e-campus project - a deployment of over 100 public displays at Lancaster that has been explicitly built as a research testbed for w ork on public display systems. The system includes support for interacti on with displays using mobile devices. As it stands e-campus is not a gl obal display network but it is in daily use and the applications and inf rastructure developed provide\, we believe\, some insight into future Gl obal Display Networks.\n \nBIO:\nProfessor Nigel Davies holds a BSc and PhD in Computer Science\, both from Lancaster University\, UK. Having co mpleted his studies he was a visiting researcher at the Swedish Institut e of Computer Science (SICS) before returning to Lancaster in 1994 to he lp create the University's Mobile Computing group. He has since managed over 3 million pounds worth of projects at Lancaster\, including the MOS T\, GUIDE and e-Campus projects\, which have been widely reported on in the academic literature and the popular press. During 1999/ 2000 he spen t a year as a visiting researcher at Sony's Distributed Systems Lab in S an Jose working on integrating mobile devices with home AV networks. In recognition of his work in establishing Lancaster as a major research ce nter in the field of mobile computing he was awarded a personal chair in the Computing Department in 2000. He has participated actively in the m obile computing research community and has served in a number of roles i ncluding Program Chair for IEEE WMCSA 2000\, Program Chair for Ubicomp 2 004 and Program Co-Chair for MobiSys 2006. In 2007 Nigel was a visiting research at both the Bonn Institute of Technology and ETH Zurich where h e worked on public display systems. Nigel was one of the first associate editors for IEEE Transactions on Mobile Computing and was a founding as sociate editor-in-chief of IEEE Pervasive Computing. From 2001-2008 he was an Associate/Adjunct Associate Professor at the University of Arizon a. He is currently Head of the Computing Department at Lancaster Univers ity.\n \nVisitor Host: M. Satyanarayanan\nVisitor Coordinator: Tracy Far bacher\, tracyf@cs.cmu.edu\n \nSDI / LCS Seminar Questions?\nKaren Linde nfelser\, karen@ece.cmu.edu\, or visit www.pdl.cmu.edu/SDI/ UID:69268E9F-3B3F-48A4-9256-BA55CCF6869B TRANSP:OPAQUE URL;VALUE=URI:message:%3C3CA15A3F637544D38FDCACD845FA3B1B@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20091022T120000 DTSTAMP:20091022T151058Z SUMMARY:SDI / LCS Seminar - Building a Global Display Network\, Nigel Da vies\, Lancaster University - UK CREATED:20091019T145639Z DTEND;TZID=US/Eastern:20091022T130000 LOCATION:GHC 8102 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:932253BC-6C84-4249-B8C1-3434989C9955 DTSTART;TZID=US/Eastern:20091020T120000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5523.html DTSTAMP:20091019T150916Z SUMMARY:Intelligence Seminar - Modeling Pandemic Influenza: Computation and Simulation of Epidemics and Other Dynamic Public Health Processes\, Donald S. Burke\, Dean\, Graduate School of Public Health and Associate Vice Chancellor for Global Health\nUniversity of Pittsburgh CREATED:20091019T150815Z DTEND;TZID=US/Eastern:20091020T130000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:F821E549-A1E0-4604-8B81-5BE99506D2A2 DTSTART;TZID=US/Eastern:20091021T160000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5516.html DTSTAMP:20091019T151102Z SUMMARY:Google Pittsburgh Talk - Statistical Language Detection in Web P ages\, Dick Sites\, Senior Staff Engineer CREATED:20091019T150956Z DTEND;TZID=US/Eastern:20091021T170000 LOCATION:Lower Level CIC END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:F23B1F25-ABA6-42F7-85BA-B6B406061CD7 DTSTART;TZID=US/Eastern:20091021T120000 URL;VALUE=URI:message:%3C001c01ca4e96$2f1f45c0$8d5dd140$@cmu.edu%3E DTSTAMP:20091019T151235Z SUMMARY:CyLab Seminar - Online Social Networks: Research Challenges and Some Results\, Peter Marbach CREATED:20091019T151235Z DTEND;TZID=US/Eastern:20091021T130000 LOCATION:CIC - INI DEC END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:6672BE2D-A6A3-4705-96BE-6C64FAA592D7 DTSTART;TZID=US/Eastern:20091023T110000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5529.html DTSTAMP:20091019T151508Z SUMMARY:Joint CMU-Pitt Ph.D. Program in \nComputational Biology Seminar - From Prediction of Structure to Design of Function\, David Baker\, \nP rofessor of Biochemistry\nUniversity of Washington CREATED:20091019T151412Z DTEND;TZID=US/Eastern:20091023T120000 LOCATION:NSH 1305 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:590B94BD-D2E8-4227-B5D2-E32C2E09CC4C DTSTART;TZID=US/Eastern:20091027T120000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5527.html DTSTAMP:20091019T151635Z SUMMARY:Special ECE Seminar -The State of Parallel Programming\, Burton J. Smith\, Technical Fellow\, Microsoft CREATED:20091019T151548Z DTEND;TZID=US/Eastern:20091027T130000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:OPAQUE UID:C1B9F437-270B-437B-861A-E4DF2194C636 DTSTART;TZID=US/Eastern:20091106T110000 URL;VALUE=URI:http://calendar.cs.cmu.edu/scsEvents/demo/5490.html DTSTAMP:20091019T152505Z SUMMARY:Joint CMU-Pitt Ph.D. Program in Computational Biology Seminar - ChIP-sequencing: Data Analysis and Applications\, Peter J. Park \, \nAss istant Professor Children's Hospital Informatics Program\nHarvard-MIT Di vision of Health Sciences and Technology CREATED:20091019T152324Z DTEND;TZID=US/Eastern:20091106T120000 LOCATION:BST3\, Room 6014 Pitt Campus END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:FF612AE8-6778-44ED-8AEE-29EB2398C59C DTSTART;TZID=US/Eastern:20091026T120000 URL;VALUE=URI:message:%3C007f01ca53ef$b8d1b240$2a7516c0$@cmu.edu%3E DTSTAMP:20091023T151548Z SUMMARY:CyLab Seminar - Starting Over After a Lost Decade\; In Search of a Bold New Vision for Cyber Security\, Richard Power CREATED:20091023T151548Z DTEND;TZID=US/Eastern:20091026T130000 LOCATION:INI DEC CIC END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\nSimulation has proven to be an invaluable tool fo r the networking\ncommunity\, supporting the controlled and systematic a ssessment of the\nimpacts of wireless environments on existing communica tion mechanisms\nand the evaluation of solutions to emerging issues. How ever\, due to\ntheir focus on networking aspects\, available modern netw ork simulators\nsimplify and abstract the details of the physical layer and the wireless\nchannel\, which can lead to the problem of not capturi ng all significant\neffects one would observe in reality. Furthermore\, the abstraction of\nthe physical layer prohibits the capability to evalu ate the impact of\nadvanced signal processing techniques on higher layer s\, in particular\nthe impact of techniques dealing with channel fading and interference\nchallenges. To address these issues\, ideas such as wi reless network\nemulator testbeds (e.g. the one at CMU Pittsburgh) or so ftware-defined\nradios (e.g. USRP\, WARP\, SORA) have been proposed rece ntly. They either\nallow the controlled and systematic evaluation of wir eless channel\nimpacts on network and protocol performance by emulating the radio\npropagation characteristics or the study of physical layer al gorithms in\nreal environments by emulating wireless communication techn ology in\nsoftware. Despite the huge success and popularity of those ide as in the\npast\, it is yet not possible to apply them to large networks or networks\nwith special characteristics. For instance when studying t he scalability\nchallenges of vehicular networks based on IEEE 802.11p c ommunication\ntechnology\, in which it is very likely that more than 100 vehicles have\nto share a common channel\, it is either very expensive to perform\nfield-operational tests using software-defined radios\, or t echnically\ndifficult (and probably also expensive) to build a wireless network\nemulator that is capable to support such a huge number of netwo rk nodes\nsimultaneously. In view of bridging the gap between physical l ayer\nemulation and network simulation\, this talk will discuss the comb ination\nof physical layer and wireless channel emulation for OFDM-based IEEE\n802.11 communication technology as well as the exemplary integrat ion of\nsuch emulation into the popular NS-3 network simulator. In parti cular\,\nthis talk discusses the major design and implementation aspects of this\napproach and the challenge of integrating a physical layer emu lator into\na time-discrete and event-based network simulator. UID:B7A1F66F-D087-462D-9E97-AAC047C107C8 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AE5FA71.7060202@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091027T150000 DTSTAMP:20091026T201403Z SUMMARY:Special seminar: Bridging the Gap between Physical Layer Emulati on and Network Simulation\, Jens Mittag\, Karlsruhe Institute of Technol ogy (KIT) CREATED:20091026T201403Z DTEND;TZID=US/Eastern:20091027T160000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:63790581-C69D-46CD-88A2-71C00A601A3D DTSTART;TZID=US/Eastern:20091104T120000 URL;VALUE=URI:message:%3C4AE87538.4060907@cs.cmu.edu%3E DTSTAMP:20091028T172144Z SUMMARY:Speaking Skills Talk - Optimally Starting and Spreading Rumors i n Social Networks\, Brendan R. Meeder CREATED:20091028T172144Z DTEND;TZID=US/Eastern:20091104T130000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:8E422FBD-ECFB-4312-A881-1D68A336D0FE DTSTART;TZID=US/Eastern:20091103T153000 DTSTAMP:20091102T205431Z SUMMARY: Intelligence Seminar \nHolger H. Hoos\, \nAssociate Professor o f Computer Science\nUniversity of British Columbia\nTaming the Complexit y Monster CREATED:20091102T205407Z DTEND;TZID=US/Eastern:20091103T163000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:E0A776FE-10FA-4E51-9D00-FAE59A40A664 DTSTART;TZID=US/Eastern:20091104T093000 DTSTAMP:20091102T205848Z SUMMARY:Sesame Street Coming to Campus! CREATED:20091102T205450Z DTEND;TZID=US/Eastern:20091104T103000 LOCATION:Rashid Auditorium (GHC 4401) END:VEVENT BEGIN:VEVENT SEQUENCE:3 TRANSP:OPAQUE UID:AAB03E9F-43A4-4F45-9D16-943F73D5A0B1 DTSTART;TZID=US/Eastern:20091105T160000 DTSTAMP:20091102T205645Z SUMMARY: Center for the Neural Basis of Cognition Colloquium \nLizabeth M. Romanski\, Associate Professor\nDepartment of Neurobiology and Anatom y\nUniversity of Rochester Medical Center\nConvergence of Faces and Voca lizations\nin the Frontal Lobe CREATED:20091102T205618Z DTEND;TZID=US/Eastern:20091105T170000 LOCATION:Biomedical Science Tower 3 (BST3) 6014\nPitt Campus END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:F5C2CF90-9600-4561-B9E1-418B26630ADC DTSTART;TZID=US/Eastern:20091106T120000 DTSTAMP:20091102T210041Z SUMMARY:Speaking Skills Talk — Jennifer Denise Tam\, \nComputer Science Department\nNo Authorization Without Representation: \nConveying the aut hority applications require\nusers to grant as a condition of installati on CREATED:20091102T205954Z DTEND;TZID=US/Eastern:20091106T130000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:DB3D2600-06CA-415F-9F97-2A4979319874 DTSTART;TZID=US/Eastern:20091103T120000 URL;VALUE=URI:message:%3C4AF03866.3030900@cs.cmu.edu%3E DTSTAMP:20091103T152400Z SUMMARY:Computer Systems Lunch today CREATED:20091103T152400Z DTEND;TZID=US/Eastern:20091103T130000 LOCATION:GHC 6121 END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:Abstract:\n\nConsider the following variant of the set-cover problem: There is a universe of elements and a collection of sets\, wit h each set S also having an associated covering requirement of K(S). The goal is to output an ordering of the elements such that the total "cove r time" of all the sets is minimized\, where the cover time of a set is the first time when K(S) elements from S have been output.\n\nThe proble m was introduced by Azar\, Gamzu\, and Yin as a theoretical modeling of a web-search ranking problem\, where the requirement is to compute an or dering of web-pages for a given search query such that the average user finds the page(s) he/she is looking for quickly\; they also gave a logar ithmic approximation algorithm. In this talk\, we will go over a simple randomized constant factor approximation algorithm for this problem.\n\n (joint work with Nikhil Bansal and Anupam Gupta)\n\nPresented in Partial Fulfillment of the CSD Speaking Skills Requirement. UID:581839F1-D234-4369-868E-859E507B7DF5 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AF2F8D5.3090707@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091111T120000 DTSTAMP:20091105T161801Z SUMMARY:Speaking Skills Talk - Approximating Generalized Min Sum Set Cov er\, Ravishankar Krishnaswamy CREATED:20091105T161722Z DTEND;TZID=US/Eastern:20091111T130000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION: http://seminar.living-environments.net/\n \n \n JIM MORRIS\ , Carnegie Mellon Silicon Valley\n LORRIE CRANOR\, Institute for Softwar e Research & Dept. of Engineering & Public Policy\n KUSAT OZENC\, School of Design UID:6C24E3AA-30EE-4E6B-8640-C323F435767D TRANSP:OPAQUE URL;VALUE=URI:message:%3C327FCE08D1971642BF5CF99E2EB20D5A02F0CDE85D@PGH- MSGCMS-01.andrew.ad.cmu.edu%3E DTSTART;TZID=US/Eastern:20091106T150000 DTSTAMP:20091105T163926Z SUMMARY:Sustainability & Computer Science Seminar: Why People Don't Want to Share Rides and What We Might Do To Change That CREATED:20091105T163926Z DTEND;TZID=US/Eastern:20091106T160000 LOCATION:Rashid Auditorium (GHC 4401) END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:Work on Java began more than 18 years ago. In that time\, J ava has achieved widespread popularity and deployment for a large and va ried set of uses. In this general talk\, we look at the history of Java "in the wild" and on some unusual ways in which Java has come to be use d. We also reflect on that history as prelude to the future\, including future directions such as deployment in mobile devices and other mobile applications. \n\nBio:\nJames Gosling received a BSc in Computer Scienc e from the University of Calgary\, Canada in 1977. He received a Ph.D. i n Computer Science from Carnegie-Mellon University in 1983. The title of his thesis was The Algebraic Manipulation of Constraints. He has built satellite data acquisition systems\, a multiprocessor version of Unix\, several compilers\, mail systems and window managers. He has also built a WYSIWYG text editor\, a constraint based drawing editor and a text edi tor called 'Emacs' for Unix systems. At Sun his early activity was as le ad engineer of the NeWS window system. He did the original design of the Java programming language and implemented its original compiler and vir tual machine. In February 2007\, James was named an officer of the Order of Canada. UID:D949CBDB-B7B3-4A67-A001-1B180DAA2FFF TRANSP:OPAQUE URL;VALUE=URI:message:%3C327FCE08D1971642BF5CF99E2EB20D5A02F0C8FAB7@PGH- MSGCMS-01.andrew.ad.cmu.edu%3E DTSTART;TZID=US/Eastern:20091106T170000 DTSTAMP:20091105T164204Z SUMMARY:THE ANDREW 25 (plus 1) Reunion: Public Talk —Reflections on Java \, the Future\, and Mobile Applications\, JAMES GOSLING — Inventor of Ja va\, Vice President and Sun Fellow Sun Microsystems CREATED:20091105T164149Z DTEND;TZID=US/Eastern:20091106T180000 LOCATION:Rashid Auditorium / GHC 4401 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nMixed integer linear programming (MILP) is a po werful representation often used to formulate decision-making problems u nder uncertainty. However\, it lacks a natural mechanism to reason about objects\, groups of objects\, and relations. First-order logic (FOL)\, on the other hand\, excels at reasoning about groups of objects\, but la cks a natural representation of uncertainty. While representing proposit ional logic in MILP has been extensively explored\, no theory exists yet for fully combining FOL with MILP. We propose a new representation\, ca lled first-order programming or FOP\, which fully subsumes both FOL and MILP. We establish formal methods for reasoning about first order progra ms\, including a sound and complete lifted inference procedure based on Gomory cuts. Since FOP can offer exponential savings in representation s ize compared to FOL\, and since our inference procedure can directly dup licate any FOL resolution proof\, we anticipate that inference in FOP wi ll be more tractable than inference in FOL for corresponding problems.\n \nPresented in Partial Fulfillment of the CSD Speaking Skills Requiremen t.\n UID:02D212BB-F9C6-4E49-A06A-1ECFA2CD7AAD TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AF32E76.7050907@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091116T120000 DTSTAMP:20091105T201254Z SUMMARY:Speaking Skills Talk - First-Order Mixed Integer Linear Programm ing\, Sue Ann Hong CREATED:20091105T201254Z DTEND;TZID=US/Eastern:20091116T130000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:3 DESCRIPTION:Abstract:\n\nRobots are often given tasks to assist humans i n their environments. A task-driven robot communicates with humans to cl arify its state and actions. The humans are often experts or designated supervisors of the robot and\, while they provide very accurate response s to the robot\, it is expensive to maintain such a pairing of robots an d humans. Instead\, we propose a more symbiotic human-robot relationship \, in which humans are benefited by the robots autonomous actions and th e robot can ask any humans in the environment for help to complete its t asks. While the symbiotic relationship can be mutually beneficial\, it c an also be a detriment reducing the usability as well as the accuracy o f the human help.\n\nMy work focuses on improving the usability of such task-driven robots that employ symbiotic human-robot interaction\, while still aiming to improve the accuracy of non-supervisor responses. Speci fically\, through varying the timing of the robots questions and the inf ormation it provides humans about its state while asking for help. We mo del both the information needs of the non-supervisor to answer a robots questions accurately as well as the perceived utility of a symbiotic rob ot in terms of the value of the tasks and cost of answering questions. W e present user studies to evaluate the usability of a symbiotic robot th at 1) provides helpful information to humans with the questions it asks and 2) uses the expected utility to determine when to ask and which acti ons to take. We conclude with current work implementing the results on a functional robot.\n\n\nPresented in Partial Fulfillment of the CSD Spea king Skills Requirement. UID:7724AED1-8DD8-4257-B880-21C944BE2F26 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AF32F43.10009@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091117T120000 DTSTAMP:20091109T163822Z SUMMARY:Speaking Skills Talk - Usability of Robots that Ask for Help\, S tephanie L. Rosenthal CREATED:20091105T201334Z DTEND;TZID=US/Eastern:20091117T130000 LOCATION:GHC 4303 END:VEVENT BEGIN:VEVENT SEQUENCE:2 DESCRIPTION:Abstract:\n\nIn machine learning\, a conjunction is the "AND " of some subset of features\; a halfspace is a linear threshold functio n on all the features.\n\nIn this talk\, we will focus on the computatio nal aspect of learning halfspaces and conjunctions. We prove the followi ng strong (computational) hardness result for learning: Given a distribu tion on labeled examples from the hypercube such that there exists a con junction consistent with (1 - eps)-fraction of the examples\, it is NP-h ard to find a halfspace that is correct on ( 1/2 +eps)-fraction of the e xamples\, for arbitrary constant eps > 0. In learning theory terms\, wea k\nagnostic learning of monomials by halfspaces is NP-hard. This hardnes s result bridges between and subsumes two previous results which showed\ nsimilar hardness results for the hardness of proper learning of\nmonomi als and halfspaces. As immediate corollaries of our result\, we give the first optimal hardness results for weak agnostic learning of decision l ists and majorities.\n\nKey techniques in the proof include sparse appro ximation of halfspace functions and invariance principles.\n\nNo theory background is assumed in order to understand the talk.\n\nPresented in P artial Fulfillment of the CSD Speaking Skills Requirement.ulfillment of the CSD Speaking Skills Requirement. UID:F3DF28F8-389A-4BDA-B194-E893D85DC2FF TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AF33110.1050103@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091118T120000 DTSTAMP:20091106T161626Z SUMMARY:Speaking Skills Talk - Hardness Result of Learning Conjunctions and Halfpsaces\, Yi Wu CREATED:20091105T201417Z DTEND;TZID=US/Eastern:20091118T130000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nAlthough C is a fairly old language\, today it still remains as one of the most used languages with a significant amoun t of code written that is being utilized by modern software and modern l anguages. Unfortunately\, C is an unsafe language that gives users a gre at deal of flexibility in terms of memory management and does little to enforce the correct use of said memory. Furthermore\, the continuous inc rease in code size and complexity and the growing number of untrusted so urces make it essential to be able to ensure code safety.\n\nOur goal is to create a C compiler that adds runtime checks to the code to enforce memory safety and certifies that its output is type safe. This compiler should generate code that runs without major overhead and without raisin g exceptions whenever its input is legal.\n\nPresented in Partial Fulfil lment of the CSD Speaking Skills Requirement.\n UID:A509CD2A-34A4-4354-AD22-624993CB767D TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AF3344D.80900@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091123T160000 DTSTAMP:20091105T202715Z SUMMARY:Speaking Skills Talk - Towards certifying compilation for C\, Mi guel Silva CREATED:20091105T202715Z DTEND;TZID=US/Eastern:20091123T170000 LOCATION:GHC 8102 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Talk Abstract\n\nLicense agreements and privacy statements a re common features of software and software services\, but less than 2% of the population actually read them. While many companies have little m otivation to compel users to read such agreements\, there are nonetheles s times when it is advantageous to effectively communicate legal terms t o one's user base. In this talk\, I present Textured Agreements\, a visu ally redesigned software agreement that captures people's attention\, co nveys the personal relevance of the content\, and increases the ease wit h which one can navigate and read the agreement. Results from experiment al studies show that this design significantly increases reading time\, leads to increased comprehension\, and is preferred by users. Furthermor e\, our data indicate these results are not simply due to the novelty of the design\, but\, rather\, due to the particular set of design element s chosen. UID:E771EABA-1CC8-4D16-A641-C55514A67FA2 TRANSP:OPAQUE URL;VALUE=URI:message:%3C005501ca6150$e4ade500$ae09af00$@cmu.edu%3E DTSTART;TZID=US/Eastern:20091109T120000 DTSTAMP:20091109T152849Z SUMMARY:CyLab Seminar — "i agreed to what?!" re-envisioning license agre ements and privacy statements. Michael Terry CREATED:20091109T152849Z DTEND;TZID=US/Eastern:20091109T130000 LOCATION:INI Distributed Education Center (DEC) on the Lobby level of th e CIC building END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:TRANSPARENT UID:307B7A73-165A-4EF1-8B86-456B54FD2037 DTSTART;VALUE=DATE:20091109 DTSTAMP:20091109T152959Z SUMMARY:PDL 17th Annual Workshop & Retreat CREATED:20091109T152929Z DTEND;VALUE=DATE:20091112 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nHybrid systems are models for complex physical sy stems and are\ndefined as dynamical systems with interacting discrete tr ansitions\nand continuous evolutions along differential equations. They\ narise frequently in many application domains\, including\naviation\, au tomotive\, railway\, and robotics. As a theoretical and\npractical found ation for deductive verification of hybrid systems\, we\npresent a dynam ic logic for hybrid systems: differential dynamic logic.\nOur verificati on approach for this logic is a compositional proof\ncalculus combining results from symbolic logic\, real algebraic\ngeometry\, differential al gebra\, and computer algebra. Our main result\nproves that this calculus axiomatizes the transition behavior of\nhybrid systems completely relat ive to differential equations. The\napproach is implemented in the hybri d theorem prover KeYmaera\, with\nwhich we have verified several interes ting case studies from ground and\nair transportation. UID:58A67C91-36D8-43AA-85C7-8AA8B6753AD3 TRANSP:OPAQUE URL;VALUE=URI:message:%3C29828_1257522975_nA6FuAPV010043_4AF44696.703040 3@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091120T153000 DTSTAMP:20091109T200209Z SUMMARY:POP Seminar — Logical Analysis of Hybrid Systems\, Andre Platzer \, Computer Science Department\, Carnegie Mellon CREATED:20091109T200209Z DTEND;TZID=US/Eastern:20091120T170000 LOCATION:GHC 9115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nMany interesting theoretical problems arise fro m computer networks. In this thesis we will consider three of them: algo rithms and data structures for problems involving distances in networks (in particular compact routing schemes\, distance labels\, and distance oracles)\, algorithms for wireless capacity and scheduling problems\, an d algorithms for optimizing iBGP overlays in autonomous systems on the I nternet. While at first glance these problems may seem extremely differe nt\, they are similar in that they all attempt to look at a previously s tudied networking problem in new\, more realistic frameworks. In other w ords\, they are all as much about new models for old problems as they ar e about new algorithms. In this thesis we will define these models\, des ign algorithms for them\, and prove hardness and impossibility results f or these three types of problems.\n\nThesis Committee:\nAnupam Gupta\, C hair\nAvrim Blum\nBruce Maggs\nMatthew Andrews\, Alcatel-Lucent Bell Lab s\n\nThesis Summary: http://www.cs.cmu.edu/~mdinitz/proposal.pdf UID:2B6F465E-5D74-4540-976B-2CC307D33376 TRANSP:OPAQUE URL;VALUE=URI:message:%3C664_1257860008_nAADXMkP000187_4AF96B53.3060401@ cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091117T140000 DTSTAMP:20091110T143847Z SUMMARY:Thesis Proposal - Algorithms and Models for Problems in Networki ng\, Michael Dinitz CREATED:20091110T143847Z DTEND;TZID=US/Eastern:20091117T150000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nThe general problem of planning for uncertain d omains remains a difficult challenge. Research that focuses on construct ing plans by reasoning with explicit models of uncertainty has produced some promising mechanisms\, but suffers from scaling difficulties. Alter nate research in robust planning has emphasized the use of deterministic planning techniques\, with the goal of scalably constructing a flexible plan that can absorb deviations during execution\, but this either resu lts in overly conservative plans\, or ignores the potential leverage tha t can be provided by explicit uncertainty models.\n\nOur approach\, call ed Probabilistic Plan Management (PPM)\, couples the strengths of both o f the above approaches\, taking advantage of the known uncertainty model while avoiding the overhead of non-deterministic planning. PPM takes as its starting point a deterministic plan that is built with deterministi c modeling assumptions\, and begins by layering an uncertainty analysis on top of the plan which calculates the overall expected outcome of exec ution and can be used to identify expected weak areas of the schedule.\n \nPPM uses the analysis in two main ways to maximize the utility of and manage execution. First\, it makes deterministic plans more robust by mi nimizing the negative impact that unexpected or undesirable contingencie s can have on plan utility. In experiments\, probabilistic schedule stre ngthening is able to significantly increase the utility of execution whi le introducing only a modest overhead.\n\nSecond\, PPM reduces the amoun t of replanning that occurs during execution via a probabilistic meta-le vel control algorithm. It uses the probability analysis as a basis for i dentifying cases where replanning probably is (or is not) necessary\, an d acts accordingly. Experiments show that probabilistic meta-level contr ol is able to considerably decrease the amount of time spent managing pl an execution\, without affecting how much utility is earned.\n\nThesis C ommittee:\nReid Simmons\, Chair\nStephen F. Smith\nManuela Veloso\nDavid J. Musliner\, SIFT\n\n UID:C2ECD953-9389-4BA8-A826-3D168DBCECE3 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4AF99308.6060500@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091117T160000 DTSTAMP:20091110T162850Z SUMMARY:Thesis Oral - Probabilistic Plan Management\, Laura M. Hiatt CREATED:20091110T162850Z DTEND;TZID=US/Eastern:20091117T170000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:ABSTRACT:\nCloud Storage is a new business model for deliver ing storage to customers\, but there are some challenges that research c ould help solve. This talk will cover:\n \nCloud Storage APIs and Standa rds (CDMI) Discussion and Brainstorming on Cloud Storage research areas such as:\n \n• performance aspects of public and private clouds\n• drivi ng the economics of cloud through management automation\n• Self-* storag e clouds\n• Clouds for media files\n• Scaling out storage clouds\n \nBIO :\nMark A. Carlson is a software engineer known in the systems managemen t industry for his work in management standards and technology. Mark was the first employee of a small startup in Boulder\, Colorado called Redc ape Policy Software. Sun Microsystems acquired the company and its techn ology in 1998 and subsequently promoted it as Jiro\, a common management framework based on Java and Jini.\n \nCarlson is probably best known fo r his work on the development of a storage management standard called SM I-S for the SNIA\, serving as the chair of the group overseeing the spec ification for several years. The specification is now an ANSI and ISO st andard.\n \nIn addition to SMI-S\, Mark also has led the development of a reference implementation of the XAM standard\, a next generation stora ge interface with support for metadata\, query and compliance based data retention of fixed content. Based on this work\, he authored the Storag e Industry Resource Domain Model\, a model for Computer data storage int erfaces showing the role of system metadata for future integrated Data S ervices orchestrated by policy.\n \nCarlson has led various efforts arou nd Policy Based Management\, including co-authoring RFC 3198 and chairin g of the DMTF and SNIA policy working groups. He is a core contributor o n the Apache incubator project called Imperius[4]\, which is an implemen tation of the standard CIM-SPL policy language.\nMark's most recent effo rts have been to lead the development of a standard Cloud Storage interf ace leveraging the preceding storage and data management technologies. A s chair of the SNIA Cloud Storage Technical Work Group\, he is driving t he development of the Cloud Data Management Interface (CDMI) and a refer ence implementation of the interface.\n \nHe is currently serving on the SNIA Technical Council.[5] He is also on the Board of Directors for the Distributed Management Task Force (DMTF)\, where he serves as the VP of Alliances.\n \nSDI / LCS Seminar Questions?\nKaren Lindenfelser\, 86716 \, or visit www.pdl.cmu.edu/SDI/ UID:EE887ADE-EB61-4727-8281-56A7E7A34209 TRANSP:OPAQUE URL;VALUE=URI:message:%3CBC4DFA16551B43359F2CC47AC2E6ABAB@pdl.cmu.edu%3E DTSTART;TZID=US/Eastern:20091112T130000 DTSTAMP:20091112T161257Z SUMMARY:SDI / LCS Seminar - Cloud Storage Standards and Research Possibi lities\, Mark Carlson\, Sun Microsystems CREATED:20091112T161257Z DTEND;TZID=US/Eastern:20091112T140000 LOCATION:CIC 2101 END:VEVENT BEGIN:VEVENT SEQUENCE:1 TRANSP:OPAQUE UID:27E11A6B-5B34-4236-927B-670A84832C3A DTSTART;TZID=US/Eastern:20091216T090000 URL;VALUE=URI:message:%3CC7219AD2.24742%25dhyde@cs.cmu.edu%3E DTSTAMP:20091112T181329Z SUMMARY:SystemStorm CREATED:20091112T181329Z DTEND;TZID=US/Eastern:20091216T100000 LOCATION:PAA END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:F776AA96-F5EB-4BE3-BF5F-77C45335B09A DTSTART;TZID=US/Eastern:20091217T093000 DTSTAMP:20091113T161951Z SUMMARY:Black Friday - Systems CREATED:20091113T161911Z DTEND;TZID=US/Eastern:20091217T120000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:5 TRANSP:OPAQUE UID:4F11F098-2145-4944-A136-9C615BCE63AB DTSTART;TZID=US/Eastern:20091217T130000 DTSTAMP:20091113T162028Z SUMMARY:Black Friday - AI / Theory CREATED:20091113T161957Z DTEND;TZID=US/Eastern:20091217T170000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:6 TRANSP:TRANSPARENT UID:3DA7B2F2-61DA-4961-91DF-5B4DB3EE4005 DTSTART;VALUE=DATE:20091217 DTSTAMP:20091113T162131Z SUMMARY:Black Friday CREATED:20091113T162058Z DTEND;VALUE=DATE:20091219 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:4 TRANSP:OPAQUE UID:AB8655C2-9C09-4137-ABB5-F9D309240388 DTSTART;TZID=US/Eastern:20091218T220000 DTSTAMP:20091113T162153Z SUMMARY:Black Friday - General Meeting CREATED:20091113T162132Z DTEND;TZID=US/Eastern:20091218T230000 LOCATION:GHC 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Monday UID:41608ECF-494A-40C9-B674-5B824B7F453E TRANSP:OPAQUE URL;VALUE=URI:message:%3C4B01809C.4040607@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091116T150000 DTSTAMP:20091116T170041Z SUMMARY:SPECIAL SCS INDUSTRY LEADER TALK: Jim Whitehurst\, President/CEO \, RedHat\, Inc. CREATED:20091116T170041Z DTEND;TZID=US/Eastern:20091116T160000 LOCATION:ASA Room/Gates & Hillman Centers 6115 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Abstract:\n\nTime series data arise in numerous applications \, such as motion capture\, computer network monitoring\, data center mo nitoring\, environmental monitoring and many more. Finding patterns in s uch sequences is crucial to understanding the data\, for example\, to bu ild humanoid robots\, to detect pollution in drinking water\, and to ide ntify intrusion in computer networks. The central theme of our work is t o answer the question: how to find interesting and unexpected patterns i n large time series?\n\nIn this proposal\, we focus on fast algorithms o n mining large collections of coevolving time series\, with or without m issing values. We will present three pieces of our work: natural stitchi ng of human motions\, time series mining and summarization with missing values\, and a parallel learning algorithm for the underlying model\, Li near Dynamical Systems (LDS). Our algorithms can mine meaningful pattern s effectively and efficiently. With those patterns\, our algorithms can perform forecasting\, compression\, and segmentation for co-evolving tim e series\, even with missing values. Furthermore\, we apply our algorith ms to solve practical problems including occlusions in motion capture\, and generating natural human motions by stitching together carefully cho sen pairs of motions. We also proposed a parallel learning algorithm for LDS to fully utilize the power of multicore/multiprocessors\, which wil l serve as corner stone of many applications and algorithms for time ser ies. All our algorithms scale linearly with respect to the length of seq uences\, and outperform the competitors often by large factors.\n\nBased on our current work\, we propose to attack some interesting problems in mining time series data\, categorized into two classes: (a) without mis sing values: including compact feature extraction\, indexing\, clusterin g and data stream monitoring\; (b) with missing values: mining under dom ain constraints\, like bone-length constraints in motion capture sequenc es. Our potential algorithms can be used to find similar motions from a database quickly\, to detect anomalies\, and to recover motion capture o cclusion.\n\nThesis Committee:\nChristos Faloutsos\, Chair\nNancy Pollar d\nEric Xing\nJiawei Han\, University of Illinois at Urbana-Champaign\n\ nThesis Summary: http://www.cs.cmu.edu/~leili/proposal/proposal.html\n UID:D1270325-E761-4396-9EAF-76BB6CDDFD59 TRANSP:OPAQUE URL;VALUE=URI:message:%3C4B06F3F0.8060502@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091201T150000 DTSTAMP:20091120T204129Z SUMMARY:Thesis Proposal - Fast Algorithms for Time Series Mining\, Lei L i CREATED:20091120T204129Z DTEND;TZID=US/Eastern:20091201T160000 LOCATION:Gates 8102 END:VEVENT BEGIN:VEVENT SEQUENCE:1 DESCRIPTION:Systems that sense\, learn\, and reason from streams of data promise to provide extraordinary value to people and society. Beyond de livering value\, harnessing theoretical principles to build systems that operate in the open world can teach us about the sufficiency of existin g models--and light the path to new research. I will discuss directions with learning and inference in the open world\, highlighting key ideas i n the context of projects in transportation\, energy\, and healthcare. F inally\, I will discuss opportunities for building systems with new kind s of open-world competencies by weaving together components that leverag e advances from several research subdisciplines.\n\n\nBIO:\nEric Horvitz is a Principal Researcher at Microsoft Research. His interests span the oretical and practical challenges with learning\, inference\, and decisi on making under uncertainty. He is a Fellow of the Association for the A dvancement of Artificial Intelligence (AAAI) and has served as President of the organization. He has also served on the NSF Computer & Informati on Science & Engineering (CISE) Advisory Board\, the DARPA Information S cience and Technology Study Group (ISAT)\, the Naval Research Advisory C ommittee (NRAC)\, and the board of the Decision Education Foundation (DE F). He received his PhD and MD degrees at Stanford University. More info rmation can be found at http://research.microsoft.com/~horvitz\n\nhttp:/ /calendar.cs.cmu.edu/dlseries/5575.html UID:2C0E85F3-677E-48AD-AC9B-EF7713D7C68E TRANSP:OPAQUE URL;VALUE=URI:message:%3C13535_1259009055_nANKi8R2020594_4B0AF3B0.801050 3@cs.cmu.edu%3E DTSTART;TZID=US/Eastern:20091203T154500 DTSTAMP:20091124T143625Z SUMMARY:BRUCE NELSON MEMORIAL LECTURE: Learning\, Inference\, and Actio n in the Open World\, Eric Horvitz CREATED:20091124T143625Z DTEND;TZID=US/Eastern:20091203T170000 LOCATION:Rashid Auditorium - GHC 4401 END:VEVENT END:VCALENDAR