Solving the small sample size problem in face recognition using generalized discriminant analysis

The goal of face recognition is to distinguish persons via their facial images. Each person's images form a cluster, and a new image is recognized by assigning it to the correct cluster. Since the images are very high-dimensional, it is necessary to reduce their dimension. Linear discriminant analysis (LDA) has been shown to be effective at dimension reduction while preserving the cluster structure of the data. It is classically defined as an optimization problem involving covariance matrices that represent the scatter within and between clusters. The requirement that one of these matrices be nonsingular restricts its application to datasets in which the dimension of the data does not exceed the sample size. For face recognition, however, the dimension typically exceeds the number of images in the database, resulting in what is referred to as the small sample size problem. Recently, the applicability of LDA has been extended by using the generalized singular value decomposition (GSVD) to circumvent the nonsingularity requirement, thus making LDA directly applicable to face recognition data. Our experiments confirm that LDA/GSVD solves the small sample size problem very effectively as compared with other current methods.     

Exploring Virtual Librarianship: Second Life Library 2.0

ABSTRACT

In April 2006, Alliance Library System and Online Programming for All Libraries partnered to start the Second Life Library 2.0 in the online world Second Life and in August 2006 purchased an island on the Teen grid of Teen Second Life. Second Life is a virtual world, a descendant of multiplayer online games. Educational institutions have recently been drawn into Second Life's robust economy and vibrant cultural life. The Second Life Library 2.0 works with librarians from around the world and with other educational partners such as TechSoup, WorldBridges, and the New Media Consortium. Second Life Library 2.0 provides “traditional” library services such as ready reference, book discussions, and search assistance, but, at heart, it explores what it means to be a virtual library in a virtual world.     

Generalizing discriminant analysis using the generalized singular value decomposition

Discriminant analysis has been used for decades to extract features that preserve class separability. It is commonly defined as an optimization problem involving covariance matrices that represent the scatter within and between clusters. The requirement that one of these matrices be nonsingular limits its application to data sets with certain relative dimensions. We examine a number of optimization criteria, and extend their applicability by using the generalized singular value decomposition to circumvent the nonsingularity requirement. The result is a generalization of discriminant analysis that can be applied even when the sample size is smaller than the dimension of the sample data. We use classification results from the reduced representation to compare the effectiveness of this approach with some alternatives, and conclude with a discussion of their relative merits.     

Effects of pH and agitation on the growth of Listeria monocytogenes Scott A in brain heart infusion broth containing combined potassium lactate and sodium diacetate during storage at 4 or 10 degrees C

The objective of this study was to compare the effects of pH on the growth kinetics of Listeria monocytogenes Scott A in static and agitated broths stored at 4 and 10 degrees C with and without a combination of 1.85% potassium lactate (PL) and 0.13% sodium diacetate (SDA) (3.3% of a 60% commercial solution, PURASAL P Opti.Form 4). The pH of brain heart infusion broth without (control) or with 1.85% PL + 0.13% SDA was adjusted to 5.5, 6.0, 6.5, and 7.5. L. monocytogenes Scott A was inoculated (at 10(2) CFU/ml) into pH-adjusted broth, which was stored at 4 or 10 degrees C with or without agitation. At pH 5.5, a listeriostatic effect was observed for the broth containing 1.85% PL + 0.13% SDA at 4 and 10 degrees C both with and without agitation. At pH 6.0, 1.85% PL + 0.13% SDA fully controlled the growth of L. monocytogenes Scott A in static broth at 4 degrees C for up to 20 days and significantly slowed the growth of the pathogen in agitated broth. At 10 degrees C, the growth of L. monocytogenes Scott A was significantly reduced by 1.85% PL + 0.13% SDA in agitated and unagitated broths. At pH 6.5, 1.85% PL + 0.13% SDA significantly suppressed the growth of L. monocytogenes Scott A at both 4 degrees C (P < 0.001) and 10 degrees C (P < 0.01). At pH 7.5, 1.85% PL + 0.13% SDA had a limited effect on the growth of L. monocytogenes Scott A in broth stored at 4 and 10 degrees C. At 4 degrees C, agitation decreased the lag time and increased the growth rate of L. monocytogenes Scott A at all tested pHs. A similar but less obvious trend was observed for broths stored at 10 degrees C. These results indicate that lactate-diacetate combinations effectively acted with low pH and temperature to inhibit the growth of L. monocytogenes Scott A.     

Effectiveness of antismoking public service announcements on children's intent to smoke.

The authors randomly assigned 5th- and 8th-grade students to 1 of 3 treatment conditions to study the effects of frequency of exposure to an antismoking public service announcement (PSA) on the students' intent to smoke over time. They found that, among younger children only, viewing an antismoking PSA at least once reduced smoking intentions, although these positive effects did not remain over time. However, the antismoking PSA was effective for younger at-risk children. After younger at-risk children viewed the antismoking PSA once, their intent to smoke decreased significantly. Notably, these positive effects remained stable over time and mirrored the smoking intentions of younger children who were not identified as at risk. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Reaction of fatty acids with N-methyl taurine

A commercially feasible process has been developed for producing salt-free fatty acyl taurates. The products are especially suited for soap-synthetic bars, contributing most importantly limesoap dispersing properties. Presented at the AOCS meeting in Los Angeles, Calif., 1959.     

"I didn't do it:" Responsibility biases in open and closed groups.

Group members often take more responsibility for the group's outcomes than others give to them. Extending evidence that egocentrism can be muted when group members are individuated (the "unpacking effect"), the authors predicted that members of open groups would be less egocentric than members of closed groups. In open groups, the possibility of membership fluctuation reduces collectiveness, breaking the group into individuals, which in turn lessens egocentrism. The authors tested these predictions in a study of groups of students working on a task relevant to their academic success. Members claimed more personal responsibility when they worked in closed groups relative to open groups (egocentrism), and lack of cohesion mediated the open- versus closed-group link to responsibility allocations. Moreover, members of open groups did not take more responsibility after a success than a failure, whereas those in closed, successful groups claimed more credit than members of failure groups (self-serving bias). The discussion suggests that open groups, although often thought to create conflict as members compete, may contribute to a reduction in both egocentrism and self-serving responsibility allocations. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Evolution of attitudes toward computers: a retrospective view

This article reviews three previous factor analysis based studies of aggregate attitudes concerning the computer milieu and compares them with a recent one done by the authors; the four studies span almost twenty years. While providing a useful basis for making generalizations about current perceptions of the computer's role, the latest study also sought to determine whether there was a group of persons who were fearful or anxious about the computer, whose concerns were masked in studies of aggregate attitudes. By using an instrument that was evaluated by an expert in phobic behaviour, it was possible to identify a populationthat was significantly different from the general population in attitude toward computers. The value of the current study, then, is two-fold. It makes possible a retrospective view of aggregate perceptions of the general population about computers and also offers a glimpse of the attributes of those computer phobic or anxious persons who have not yet adapted to computer technology.