Karen W. Brannon

Coordinating backup/recovery and data consistency between database and file systems

Managing a combined store consisting of database data and file data in a robust and consistent manner is a challenge for database systems and content management systems. In such a hybrid system, images, videos, engineering drawings, etc. are stored as files on a file server while meta-data referencing/indexing such files is created and stored in a relational database to take advantage of efficient search. In this paper we describe solutions for two potentially problematic aspects of such a data management system: backup/recovery and data consistency. We present algorithms for performing backup and recovery of the DBMS data in a coordinated fashion with the files on the file servers. Our algorithms for coordinated backup and recovery have been implemented in the IBM DB2/DataLinks product [1]. We also propose an efficient solution to the problem of maintaining consistency between the content of a file and the associated meta-data stored in the DBMS from a readers point of view without holding long duration locks on meta-data tables. In the model, an object is directly accessed and edited in-place through normal file system APIs using a reference obtained via an SQL Query on the database. To relate file modifications to meta-data updates, the user issues an update through the DBMS, and commits both file and meta-data updates together.

Crystallographic Aspects of Low Energy Boron Implantation Into Silicon

The computer program MARLOWE has been used to investigate low energy boron implantation into silicon. When implanted in a “random” direction at 5 keV, the most deeply penetrating ions are seen to have spent an appreciable part of their path travelling in , or axial directions or in {111} or {220} planar channels. Channeling was then investigated more directly by observing the mean path travelled by the ions in the direction of incidence as a function of both azimuthal and tilt angles between the direction of incidence and the 001 direction. At 5 keV, {111} planar channeling and , , and axial channeling is prominent, with {220} planar and and axial channeling observed to a lesser degree. Higher order planar channeling such as {400}, {331}, {422} or (511} was not observed at 5 keV. At 0.75 keV only . and axial channeling and {111} planar channeling remained.

Archive storage system design for long-term storage of massive amounts of data

A dramatic shift is underway in how organizations use computer storage. This shift will have a profound impact on storage system design. The requirement for storage of traditional transactional data is being supplemented by the necessity to store information for long periods. In 2005, a total of 2,700 petabytes of storage was allocated worldwide for information that required long-term retention, and this amount is expected to grow to an estimated 27,200 petabytes by 2010. In this paper, we review the requirements for long-term storage of data and describe an innovative approach for developing a highly scalable and flexible archive storage system using commercial off-the-shelf (COTS) components. Such a system is expected to be capable of preserving data for decades, providing efficient policy-based management of the data, and allowing efficient search and access to data regardless of data content or location.

Multiple Kernel Completion and its application to cardiac disease discrimination

Data is only as good as the similarity metric used to compare it. The all important notion of similarity allows us to leverage knowledge derived from prior observations to predict characteristics of new samples. In this paper we consider the problem of compiling a consistent and accurate view of similarity given its multiple incomplete and noisy approximations. We propose a new technique called Multiple Kernel Completion (MKC), which completes given similarity kernels as well as finds their best combination within a Support Vector Machine framework, so as to maximize the discrimination margin. We demonstrate the effectiveness of the proposed technique on datasets from UCI Machine Learning repository as well as for the task of heart valve disease discrimination using CW Doppler images. Our empirical results establish that MKC consistently outperforms existing data completion methods like 0-imputation, mean-imputation and matrix completion across datasets and training set sizes.

PSAR: Predictive space aggregated regression and its application in valvular heart disease classification

This paper presents a predictive space aggregated regression based boosting algorithm, and its application in classifying the Continuous Wave(CW) Flow Doppler image data set with the diseases of stenosis and regurgitation in mitral and aortic valves. The proposed algorithm involves finding a way to simultaneously combine all the weak learners based on a well-justified assumption as in the previous work[1] that not only the weak learners but each training sample should have different contributions toward learning the final strong hypothesis. However, the proposed algorithm greatly improves on the previous method by (1) dramatically reducing the number of combination weights, leading to a more stable numerical solution, (2) having regularization in both data and predictive spaces to reduce the generalization error of the model, and (3) using the sparse weight selection scheme in the testing to further avoid overfitting. A sparse subset of the training data is chosen to best approximate the test sample, and the final hypothesis is constructed based only on the chosen training samples and associated weak learner weights. Finally, we empirically show that the proposed technique not only successfully solves the overfitting problem but also significantly increases the performance of the weak classifiers via a set of comparison experiments on the CW Flow Doppler image data set consisting of 4 types of valvular diseases at different severity levels.

Secure Information Caching on the Web

Web has become the key information source over the last few years and caching has been exploited and applied to reduce web server and network congestion as well as to improve response time. There are three types of web caching techniques: forward proxy, reverse proxy, and transparent proxy. Forward proxy server caches data close to users and is targeted to improve the user/browser response time. On the contrary, reverse proxy server normally locates next to a back-end server and is mainly targeted to reduce back-end server workload. As the web moving from primarily for information sharing to also becoming a key platform for business operations, a new generation of caching mechanism is needed to enforce the security of business content and to preserve the confidentiality of personal information. In this paper, we describe a web caching system that enhances the caching function by optionally enforcing fine grain access control rules, set by the back-end servers, on the cached content. Our system takes advantage of edge (proxy) server technology for delivering data/information from locations adjacent to users while enforcing access control rule set for each piece of cached content or page fragment.