Google Tech Talks
January, 17 2008
ABSTRACT
Team Cornell was one of six teams to complete the 2007 DARPA Urban Challenge, completing over 55 miles of autonomous driving in an urban environment in approximately seven hours, including competition stops. The competition included many urban driving scenarios such as staying in a lane, merging into traffic, passing, intersections, parking, and even robot-robot interaction. Team Cornell designed and built a vehicle around technological innovations in vehicle automation, a real time UDP based data distribution system, tightly coupled pose estimation, scene estimation including localization within an urban environment and tracking all obstacles with a fusion of laser, radar and vision sensors, and hierarchical intelligent planning. Team Cornell’s vehicle was designed to drive “human-like” with smooth, intelligent behaviors, even in the presence of a vast array of uncertainties. The systematic approach taken by Team Cornell led to an innovative, robust solution to the complex problem proposed in the 2007 DARPA Urban Challenge. This seminar will present the key technologies, semi-final and final results, and plans for future research.
Speaker: Dan Huttenlocher
Dan Huttenlocher is the John P. and Rilla Neafsey Professor of Computing, Information Science and Business at Cornell University, where he holds a joint appointment in the Computer Science Department and the Johnson Graduate School of Management. His research interests are in computer vision, social and information networks, collaboration tools, geometric algorithms and financial trading systems. He has been recognized for his research and teaching contributions on several occasions, including being named an NSF Presidential Young Investigator, New York State Professor of the Year and a Fellow of the ACM. In addition to academic posts he has been chief technical officer of Intelligent Markets, a provider of advanced trading systems on Wall Street, and spent more than ten years at Xerox PARC directing work that led to the ISO JBIG2 image-compression standard.
Speaker: Mark Campbell
Mark Campbell is an Associate Professor in the Sibley School of Mechanical and Aerospace Engineering at Cornell University. His research interests are in the areas of autonomous systems, probabilistic models of human decision making, nonlinear estimation theory, cooperative vehicle control and estimation, and sensor fusion. He has been recognized from NASA for his modeling and control work on the Middeck Active Control Experiment, flown on STS-67 in 1995. He received best paper awards from the AIAA and Frontiers in Education conference, and teaching awards Cornell, University of Washington, and the ASEE. He was also an Australian Research Council International Fellowship in 2006 while on sabbatical at the University of Sydney. He is an Associate Fellow of the AIAA, an Associate Director of the AACC board, and member of the AIAA GNC Technical Committee, and is active in both IEEE and ASEE.
Duration : 1:6:12
Read the rest of this entry »
Google Tech Talk
January 21, 2010
ABSTRACT
Presented by John McCann.
We tend to think of digital imaging and the tools of Photoshop(TM) as a new phenomenon in imaging. We are also familiar with multiple-exposure HDR techniques intended to capture a wider range of scene information, than conventional film photography. We know about tone-scale adjustments to make better pictures. We tend to think of everyday, consumer, silver-halide photography as a fixed window of scene capture with a limited, standard range of response. This description of photography is certainly true, between 1950 and 2000, for instant films and negatives processed at the drugstore. These systems had fixed dynamic range and fixed tone-scale response to light. All pixels in the film have the same response to light, so the same light exposure from different pixels was rendered as the same film density.
Ansel Adams, along with Fred Archer, formulated the Zone System, starting in 1940. It was earlier than the trillions of consumer photos in the second half of the 20th century, yet it was much more sophisticated than today’s digital techniques. This talk will describe the chemical mechanisms of the zone system in the parlance of digital image processing. It will describe the Zone System’s chemical techniques for image synthesis. It also discusses dodging and burning techniques to fit the HDR scene into the LDR print. These techniques introduced spatial changes in the print causing dynamic range compression of the high-dynamic-range scene into the low-dynamic-range print. Preserving edge information is key to successful scene rendition.
Note: The recording missed the first couple minutes (the start of John’s story of Ansel’s visiting Polaroid and trying out a prototype SX-70 camera).
John McCann received a B.A. degree in Biology from Harvard University in 1964. He worked in, and later managed, the Vision Research Laboratory at Polaroid from 1961 to 1996. He has studied human color vision, digital image processing, large format instant photography and the reproduction of fine art. He is a Fellow of IS&T. He is a past President of IS&T and the Artists Foundation, Boston. He is currently consulting and continuing his research on color vision. He is the IS&T/OSA 2002 Edwin H. Land Medalist and IS&T 2005 Honorary Member and a 2008 Fellow of the Optical Society of America.
Duration : 0:59:23
Read the rest of this entry »
Google Tech Talk
December 7, 2009
ABSTRACT
CMake/CPack/CTest/CDash Open Source Tools to Build Test and Deploy C++ Software, presented by Bill Hoffman.
CMake has been in development since 1999, and has been used on several large open source projects such as ITK, VTK, ParaView, VXL, Trilinos and CMake itself. Further, KDE, one of the largest OSS projects has adopted CMake, demonstrating that CMake is capable of successfully supporting complex and large software systems. Hence CMake usage is growing rapidly with thousands of daily downloads and inclusion in several Linux distributions.
Unlike many build systems, CMake is designed to be used in conjunction with native build tools enabling developers to use makefiles, Kdevelop projects, Xcode projects, and even MS Visual Studio projects. A simple input language (included in a CMakeLists.txt file) is used to specify which files to build and what types of system introspection tests need to be performed to build the software. A persistent cache file is used to store the system information and avoid the need for user-defined environment variables.
In addition to building software, CMake provides a testing client (CTest) that integrates with the web-based CDash testing server. This server creates dashboards that build a snapshot of the software at a given time. This is critical to cross-platform development since often a change on one platform fails to compile on another one. The testing system provides for nightly builds which use a copy of the software at a specific time each night, experimental tests that can be used to share build results with other developers before committing source code, and continuous build results that test the build each time files are committed to the source control system.
Once the software is built and tested, the CPack tool can be used to package the software. CPack works similar to CMake in that it generates package information for native packaging tools. NSIS, RPM, OSX packages self extracting tar.gz, tar.gz, tar.zip can all be created. CPack information is included as a simple extension to the CMake build files.
This talk will cover the history and features of CMake, CTest, CDash and CPack in the context of a integrated development environment.
Mr. Hoffman is currently Vice President and CTO for Kitware, Inc. He is a founder of Kitware and has been part of the management team since 1999. Bill has 20 years of experience with large C++ systems. He is a lead architect of the CMake cross-platform build system and co-author of the Mastering CMake book. Mr. Hoffman is also involved in the development of the Kitware Quality Software Process and CDash, the software testing server. Mr. Hoffman developed the C++/COM wrapping technology used to create Kitware’s ActiViz product line. He has also made major contributions to VTK, ITK and ParaView. As CTO for Kitware, he guides the implementation and development of large-scale computing solutions, and oversees computer infrastructure decisions, including developing Kitware’s E-Store technology.
Mr. Hoffman received a B.S in Computer Science from the University of Central Florida, and an M.S in Computer Science from Rensselaer Polytechnic Institute (RPI). He is a graduate of the GE Software Technology Program. Prior to joining Kitware he spent nine years at GE’s Global Research Center working in the Computer Vision Group. As an expert in C++ and object-oriented programming, he has planned and taught several graduate level courses at RPI, as well as a course on object-oriented programming at New York University. Mr. Hoffman has been an invited speaker at national and international conferences, such as OSCON, FOSDEM and KDE Developer’s Conferences.
Duration : 1:8:4
Read the rest of this entry »
Google Tech Talks
May 15, 2008
ABSTRACT
Severe flood disasters may frequently occur in near future because global warming will change the flood risk. To prepare the disaster, we propose global flood simulation software which simulates flood flow in
short time on an earth viewer (i.e., global geographical information system). User can edit a flood scenario such as the location of levee failure by clicking on the earth viewer, then the software simulate the
flood flow based on fluid dynamics. The simulation and GIS communicate each other using our technology, the flood information are visualized on the viewer even during the simulation. I will make a presentation about
the technology and show how to work our flood simulation software.
Speaker: Satoshi Yamaguchi
I specialize in the geophysics, hydrodynamics and software development such as geographical information system (GIS). I conduct a research on flood disaster and develop flood simulation software on our original GIS for a number of years in Hitachi Central Research Laboratory. I studied physical oceanography in Tohoku University and received masters degree.
Duration : 0:26:12
Read the rest of this entry »
Google Tech Talks
June, 9 2008
ABSTRACT
Presented by
Andrew B. Whinston
Center for Research in Electronic Commerce
Mccombs School of Business
The University of Texas at Austin
Based on a joint research with Jianqing Chen and De Liu
Abstract:
Internet-based advertising continues to increase in importance. Internet advertising providers such as Google and Yahoo! allocate their advertising resources using a novel form of share auctions in which the highest advertiser gets the largest share, the second highest advertiser gets the second largest share, and so on. A share structure problem arises in such a setting regarding how much resources to set aside for the highest advertiser, for the second highest advertiser, etc. We address this problem under a general specification and derive implications on how the optimal share structure should change with advertisers’ price elasticity of demand for exposure, their valuation distribution, total resources, and minimum bids.
Full Synopsis here:
http://docs.google.com/a/google.com/Doc?id=cdvbh2g5_1621cg97gcct
Speaker: Dr. Andrew B. Whinston
http://en.wikipedia.org/wiki/Andrew_B._Whinston
Andrew B. Whinston is an American economist and computer scientist. He is the Hugh Roy Cullen Centennial Chair in Business Administration, Professor of Information Systems, Computer Science and Economics, and Director of the Center for Research in E-Commerce (CREC) in the McCombs School of Business at the University of Texas at Austin. Prof. Whinston is also a fellow of the IC2 Institute, Austin.
At the International Conference on Information Systems (ICIS) in December 2005, Prof. Whinston was awarded “The LEO Award for Lifetime Exceptional Achievement in Information Systems” by the Association for Information Systems (AIS).
Andrew Whinston has been a major contributor to the field since information systems began as an academic discipline. His research record is exemplary, his publications are many, his doctoral graduates are found throughout the field in all parts of the world, and his innovative research has enlightened critical developments in the field.
Duration : 1:6:7
Read the rest of this entry »
Google Tech Talk
October 23, 2009
ABSTRACT
Presented by Daniel Gruhl.
Currently, data analytics technology is in high demand as people try to extract as much value as possible from their most valuable resource – the information around them, whether in their organizations or freely and publicly available. Unfortunately, though many data analytics efforts are focused a particularly interesting (and often difficult) question, whose answer hopefully lies in the data, these projects tend to spend most of their cycles acquiring and ingesting data. Thus, the focus of these efforts tend to tilt away from data analysis and towards data ingestion. MONGOOSE is 1) A suite of technologies that one can plug domain knowledge cartridges into and that outputs data suitable for OLAP or BI consumption. One plugs in small amounts of domain knowledge that involves pulling in unstructured, semi-structured and structured data, and MONGOOSE converts it all into structured form. 2) A Platform for Worst-Case Scenario Workflow Management. MONGOOSE is built on the assumption that failure happens and it must be handled quickly and seamlessly, such that it does not stop or hinder information ingest. 3) A Platform for Community-Based Information Extraction around specific phenomenon that can be fed into statistical analysis tools.
Daniel Gruhl (dgruhl@almaden.ibm.com) is a research staff member in the Computer Science Department of IBM Almaden Research Center, San Jose, CA. Dan is currently in the Health Informatics research group. Dan specializes in very large scale text analytics for a variety of applications from healthcare to pop music. Dan co-architected IBM’s Unstructured Information Management Architecture (UIMA), which is now the de facto standard for text analytics projects. He earned his Ph.D. in electrical engineering from the Massachusetts Institute of Technology in 2000 with thesis work on distributed text analytics systems. Dan was named in MIT’s Technology Review Top 100 (TR 100) in 2004.
Varun Bhagwan (vbhagwan@us.ibm.com) is an advisory software engineer in the Computer Science Department of IBM Almaden Research Center, San Jose, CA. His interests lie in the field of text analytics, data mining, machine learning/AI, internet technologies, and services science. Since joining IBM research in 2001, Varun has worked at multiple levels of a large scale text mining project, ranging from cluster management, to indexing a multi-billion page corpus, to crawling the internet. He is currently a member of the the Health Informatics research group. Varun holds a Master’s degree in Computer Science from University of Florida, Gainesville and is currently pursuing a Ph.D. at the University of California, Santa Cruz.
Tyrone Grandison (tyroneg@us.ibm.com) manages the Intelligent Information Systems team in the Computer Science department at the IBM Almaden Research Center, San Jose, CA. Tyrone’s research interests are in data disclosure management relevant and applicable to industry verticals. Over the years, Tyrone has worked in data privacy, RFID data management, privacy-preserving mobile data management and text analytics. Tyrone is a senior member of both the ACM and IEEE and was named Pioneer of the Year by NSBE in 2009. Tyrone received a Ph.D. from Imperial College, London and M.Sc. and B.Sc. degrees from the University of the West Indies, Mona, Jamaica.
Duration : 0:41:35
Read the rest of this entry »
Google Tech Talks
October 31, 2008
ABSTRACT
In knowledge-based information retrieval, search engines consult external sources of knowledge ontologies, taxonomies, thesauri, glossaries, gazeteers to help process the documents they encounter and the requests they receive. The idea is old, obvious, and compelling but results have been singularly unimpressive. The best performing and most widely used search systems are still those that deal in lexical character patterns without using any structured knowledge to understand them.
Wikipedia is changing all that. This open, constantly evolving encyclopedia represents a vast pool of topics and semantic relations. It is arguably the largest knowledge base humanity has ever seen. At last we have a resource that is (or may be) sufficiently broad, deep, and timely to be applicable to open-domain information retrieval. However, it brings its own challenges. Wikipedia’s haphazard and only partially machine-readable structure bears little resemblance to the carefully crafted knowledge bases that have been used to assist information retrieval in the past.
This talk will discuss Wikipedia’s promises and shortcomings, and describe ongoing investigations of how best to apply it to organizing and retrieving information.
Speaker: David Milne
David Milne is a PhD student at the University of Waikato in New Zealand, where he studies under the supervision of Prof. Ian H. Witten.
Duration : 0:49:27
Read the rest of this entry »
Google Tech Talks
October 31, 2008
ABSTRACT
In knowledge-based information retrieval, search engines consult external sources of knowledge ontologies, taxonomies, thesauri, glossaries, gazeteers to help process the documents they encounter and the requests they receive. The idea is old, obvious, and compelling but results have been singularly unimpressive. The best performing and most widely used search systems are still those that deal in lexical character patterns without using any structured knowledge to understand them.
Wikipedia is changing all that. This open, constantly evolving encyclopedia represents a vast pool of topics and semantic relations. It is arguably the largest knowledge base humanity has ever seen. At last we have a resource that is (or may be) sufficiently broad, deep, and timely to be applicable to open-domain information retrieval. However, it brings its own challenges. Wikipedia’s haphazard and only partially machine-readable structure bears little resemblance to the carefully crafted knowledge bases that have been used to assist information retrieval in the past.
This talk will discuss Wikipedia’s promises and shortcomings, and describe ongoing investigations of how best to apply it to organizing and retrieving information.
Speaker: David Milne
David Milne is a PhD student at the University of Waikato in New Zealand, where he studies under the supervision of Prof. Ian H. Witten.
Duration : 0:49:27
Read the rest of this entry »
Google Tech Talks
October, 23 2007
ABSTRACT
For most of us who work on the Internet, the Web is all we have ever really known. It’s almost impossible to imagine a world without browsers, URLs and HTTP. But in the years leading up to Tim Berners-Lee’s world-changing invention, a few visionary information scientists were exploring alternative systems that often bore little resemblance to the Web as we know it today. In this presentation, author and information architect Alex Wright will explore the heritage of these almost-forgotten systems in search of promising ideas left by the historical wayside.
The presentation will focus on the pioneering work of Paul Otlet, Vannevar Bush, and Doug Engelbart, forebears of the 1960s and 1970s like Ted Nelson, Andries van Dam, and the Xerox PARC team, and more recent forays like Brown’s Intermedia system. We’ll trace the heritage of these systems and the solutions they suggest to present day Web quandaries, in hopes of finding clues to the future in the recent technological past.
Speaker: Alex Wright
Alex Wright is an information architect at the New York Times and the author of Glut: Mastering Information Through the Ages. Previously, Alex has led projects for The Long Now Foundation, California Digital Library, Harvard University, IBM, Microsoft, Rollyo and Sun Microsystems, among others. He maintains a personal Web site at http://www.alexwright.org/
Duration : 0:59:34
Read the rest of this entry »
Google Tech Talks
October, 23 2007
ABSTRACT
For most of us who work on the Internet, the Web is all we have ever really known. It’s almost impossible to imagine a world without browsers, URLs and HTTP. But in the years leading up to Tim Berners-Lee’s world-changing invention, a few visionary information scientists were exploring alternative systems that often bore little resemblance to the Web as we know it today. In this presentation, author and information architect Alex Wright will explore the heritage of these almost-forgotten systems in search of promising ideas left by the historical wayside.
The presentation will focus on the pioneering work of Paul Otlet, Vannevar Bush, and Doug Engelbart, forebears of the 1960s and 1970s like Ted Nelson, Andries van Dam, and the Xerox PARC team, and more recent forays like Brown’s Intermedia system. We’ll trace the heritage of these systems and the solutions they suggest to present day Web quandaries, in hopes of finding clues to the future in the recent technological past.
Speaker: Alex Wright
Alex Wright is an information architect at the New York Times and the author of Glut: Mastering Information Through the Ages. Previously, Alex has led projects for The Long Now Foundation, California Digital Library, Harvard University, IBM, Microsoft, Rollyo and Sun Microsystems, among others. He maintains a personal Web site at http://www.alexwright.org/
Duration : 0:59:34
Read the rest of this entry »