Conferences, Lectures, & Seminars
Events for December
-
USC CS Colloquium Lecture Series
Tue, Dec 05, 2006 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
An Zhu
GoogleTitle: Towards Achieving Anonymity Abstract:We study the problem of publishing data from a table containing personal data, in a privacy preserving manner. In particular, we aim to anonymize quasi-identifiers, i.e., non-key attributes that combinedly identifiy a unique record in the table.The first model is proposed by Sweeney, called k-anonymity. This approach suppresses some values of the quasi-identifiers, such that for every record in the modified table, there are at least k-1 other records with exactly the same value. And the quality measure here is the number of quasi-identifier values suppressed. We provide a O(k)-approximation algorithm for this problem, improving upon the previous O(k log k) result. We also show that this is the best approximation bound possible using the distance representation. For small values of k, we provide improved bounds as well.We propose a second model which generalizes the quasi-identifier values via clustering. The records are first clustered and then the cluster centers are published. To ensure privacy, we impose the constraint that each cluster must contain at least k records. We consider the measure of minimizing the maximum cluster radius, for which we provide a tight 2-approximation algorithm. The second measure concerns minimizing the sum, over all clusters, the product of number of records per cluster and the cluster radius. For this measure we also provide a constant approximation algorithm. Further, we extend the algorithms to handle the case where we can omit outliners.This talk is based on two papers:
Anonymizing Tables (ICDT 05), coauthored with Aggarwal, Feder, Kenthapadi, Motwani, Panigrahy, and Thomas.
Achieveing Anonymity via Clustering (PODS 06), coauthored with Aggrawal, Feder, Kenthapadi, Khuller, Panigrahy, and Thomas.Bio:
An obtained her phd from Stanford University in 2004, under the supervision of Rajeev Motwani and Leo Guibas. An joined Google after her graduation. Since then, An has been involved in a variety of projects at Google, including search quality/ranking, image search, scholar search, and search infrastructureLocation: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Nancy Levien
-
Robotics at Microsoft
Tue, Dec 12, 2006 @ 12:00 PM - 01:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Dr. Stewart TansleyMicrosoft ResearchTitle:
Robotics at Microsoft -- A Personal Journey & PerspectiveAbstract:
Three years ago, Microsoft's answer to roboticists was "Have you tried Windows CE? It's hard real-time you know!". Today, in Fall 2006, we have an entire dedicated software platform for robotics researchers, companies, and hobbyists, and a leading academic program to help make robotics in computer science education a potentially pervasive experience across the US and beyond. This talk will try to explain how this happened, what is available for roboticists today, and where things are going -- from the perspective of one of those at the center of these exciting developments. While very much a personal perspective, attendees at the talk will hopefully gain a better understanding of Microsoft, its relationship with academia via Microsoft Research in particular, and what it means to make a difference as one person in over 70,000.Bio:
Stewart is responsible for Embedded Systems and Robotics as part of External Research & Programs in Microsoft Research. Before this, he worked on Microsoft's production IPv6 software as part of the Windows Networking team. Prior to joining Microsoft in 2001, Stewart spent 13 years in the telecommunications industry in various technical and management positions in network software research and development, focusing on technology transfer. Stewart has a Ph.D. in Artificial Intelligence applied to Engineering from the University of Technology, Loughborough, UK. He has published a variety of papers in artificial intelligence and network management, several patents, and co-authored a book on software engineering for artificial intelligence applications.Web:
http://research.microsoft.com/~stansley/Hosted by Maja Mataric'Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Nancy Levien
