On-screen pre-deblurring of digital images using the wavefront aberration function of the human eye to improve computer access for the visually impaired

Traditional Optics has provided ways to compensate some common visual limitations (up to second order visual impairments) through spectacles or contact lenses. Recent developments in wavefront science make it possible to obtain an accurate model of the Point Spread Function (PSF) of the human eye. Through what is known as the "Wavefront Aberration Function" of the human eye, exact knowledge of the optical aberration of the human eye is possible, allowing a mathematical model of the PSF to be obtained. This model could be used to pre-compensate (inverse-filter) the images displayed on computer screens in order to counter the distortion in the user's eye. This project takes advantage of the fact that the wavefront aberration function, commonly expressed as a Zernike polynomial, can be generated from the ophthalmic prescription used to fit spectacles to a person. This allows the pre-compensation, or onscreen deblurring, to be done for various visual impairments, up to second order (commonly known as myopia, hyperopia, or astigmatism). The technique proposed towards that goal and results obtained using a lens, for which the PSF is known, that is introduced into the visual path of subjects without visual impairment will be presented. In addition to substituting the effect of spectacles or contact lenses in correcting the loworder visual limitations of the viewer, the significance of this approach is that it has the potential to address higher-order abnormalities in the eye, currently not correctable by simple means.

Design and evaluation of a consulting system for database design

Database design is a difficult problem for non-expert designers. It is desirable to assist such designers during the problem solving process by means of a knowledge based (KB) system. Although a number of prototype KB systems have been proposed, there are many shortcomings. Firstly, few have incorporated sufficient expertise in modeling relationships, particularly higher order relationships. Secondly, there does not seem to be any published empirical study that experimentally tested the effectiveness of any of these KB tools. Thirdly, problem solving behavior of non-experts, whom the systems were intended to assist, has not been one of the bases for system design. In this project, a consulting system, called CODA, for conceptual database design that addresses the above short comings was developed and empirically validated. More specifically, the CODA system incorporates (a) findings on why non-experts commit errors and (b) heuristics for modeling relationships. Two approaches to knowledge base implementation were used and compared in this project, namely system restrictiveness and decisional guidance (Silver 1990). The Restrictive system uses a proscriptive approach and limits the designer's choices at various design phases by forcing him/her to follow a specific design path. The Guidance system approach, which is less restrictive, involves providing context specific, informative and suggestive guidance throughout the design process. Both the approaches would prevent erroneous design decisions. The main objectives of the study are to evaluate (1) whether the knowledge-based system is more effective than the system without a knowledge-base and (2) which approach to knowledge implementation - whether Restrictive or Guidance - is more effective. To evaluate the effectiveness of the knowledge base itself, the systems were compared with a system that does not incorporate the expertise (Control). An experimental procedure using student subjects was used to test the effectiveness of the systems. The subjects solved a task without using the system (pre-treatment task) and another task using one of the three systems, viz. Control, Guidance or Restrictive (experimental task). Analysis of experimental task scores of those subjects who performed satisfactorily in the pre-treatment task revealed that the knowledge based approach to database design support lead to more accurate solutions than the control system. Among the two KB approaches, Guidance approach was found to lead to better performance when compared to the Control system. It was found that the subjects perceived the Restrictive system easier to use than the Guidance system.

The Effect of Authentic Literacy Experiences as Book Buddies with Hispanic Fourth Graders on Preservice Teachers’ Literacy Content Knowledge and Reading Maturity

From a sociocultural perspective, individuals learn best from contextualized experiences. In preservice teacher education, contextualized experiences include authentic literacy experiences, which include a real reader and writer and replicate real life communication. To be prepared to teach well, preservice teachers need to gain literacy content knowledge and possess reading maturity. The purpose of this study was to examine the effect of authentic literacy experiences as Book Buddies with Hispanic fourth graders on preservice teachers’ literacy content knowledge and reading maturity. The study was a pretest/posttest design conducted over 12 weeks. Preservice teacher participants, the focus of the study, were elementary education majors taking the third of four required reading courses in non-probabilistic convenience groups, 43 (n = 33 experimental, n = 10 comparison) Elementary Education majors. The Survey of Preservice Teachers’ Knowledge of Teaching and Technology (SPTKTT), specifically designed for preservice teachers majoring in elementary or early childhood education and the Reading Maturity Survey (RMS) were used in this study. Preservice teachers chose either the experimental or comparison group based on the opportunity to earn extra credit points (experimental = 30 points, comparison = 15). After exchanging introductory letters preservice teachers and Hispanic fourth graders each read four books. After reading each book preservice teachers wrote letters to their student asking higher order thinking questions. Preservice teachers received scanned copies of their student’s unedited letters via email which enabled them to see their student’s authentic answers and writing levels. A series of analyses of covariance were used to determine whether there were significant differences in the dependent variables between the experimental and comparison groups. This quasi-experimental study tested two hypotheses. Using the appropriate pretest scores as covariates for adjusting the posttest means of the subcategory Literacy Content Knowledge (LCK), of the SPTKTT and the RMS, the mean adjusted posttest scores from the experimental group and comparison group were compared. No significant differences were found on the LCK dependent variable using the .05 level of significance, which may be due to Type II error caused by the small sample size. Significant differences were found on RMS using the .05 level of significance.

The Validity of Broad and Narrow Personality Traits For Predicting Job Performance: The Differential Effects of Time

Research into the dynamicity of job performance criteria has found evidence suggesting the presence of rank-order changes to job performance scores across time as well as intraindividual trajectories in job performance scores across time. These findings have influenced a large body of research into (a) the dynamicity of validities of individual differences predictors of job performance and (b) the relationship between individual differences predictors of job performance and intraindividual trajectories of job performance. In the present dissertation, I addressed these issues within the context of the Five Factor Model of personality. The Five Factor Model is arranged hierarchically, with five broad higher-order factors subsuming a number of more narrowly tailored personality facets. Research has debated the relative merits of broad versus narrow traits for predicting job performance, but the entire body of research has addressed the issue from a static perspective -- by examining the relative magnitude of validities of global factors versus their facets. While research along these lines has been enlightening, theoretical perspectives suggest that the validities of global factors versus their facets may differ in their stability across time. Thus, research is needed to not only compare the relative magnitude of validities of global factors versus their facets at a single point in time, but also to compare the relative stability of validities of global factors versus their facets across time. Also necessary to advance cumulative knowledge concerning intraindividual performance trajectories is research into broad vs. narrow traits for predicting such trajectories. In the present dissertation, I addressed these issues using a four-year longitudinal design. The results indicated that the validities of global conscientiousness were stable across time, while the validities of conscientiousness facets were more likely to fluctuate. However, the validities of emotional stability and extraversion facets were no more likely to fluctuate across time than those of the factors. Finally, while some personality factors and facets predicted performance intercepts (i.e., performance at the first measurement occasion), my results failed to indicate a significant effect of any personality variable on performance growth. Implications for research and practice are discussed.

The Effects on Students' Self-Efficacy Beliefs Regarding Their Comprehension of American Literature When Aesthetic Reading and Reader Response Strategy are Implemented

High-stakes testing and accountability have infiltrated the education system in the United States; the top priority for all teachers must be student progress on standardized tests. This has resulted in the predominance of reading for test-taking, (efferent reading), in the English, language arts, and reading classrooms. Authentic uses of print activities, like aesthetic reading, that encourage students to engage individually with a text, have been pushed aside. During a 3-week time period, regular level, English 3/American literature students in a Title I magnet high school, participated in this quasi-experimental study (N = 62). It measured the effects of an intervention of reading American literature texts aesthetically and writing aesthetically-evoked reader responses on students’ self-efficacy beliefs regarding their comprehension of American literature. One trained teacher and the researcher participated in the study: student participants were pre- and post- tested using the Confidence in Reading American Literature Survey which examined their self-efficacy beliefs regarding their comprehension of American literature. Several statistical analyses were performed. The results of the linear regression analyses partially supported a positive relationship between aesthetically-evoked reader responses and students’ self-efficacy beliefs regarding their comprehension of American literature. Additionally, the results of the 2 (sex) x 2 (treatment) ANCOVAs conducted to test group differences in self-efficacy beliefs regarding the comprehension of American literature between treatment and control groups indicated a main effect for treatment (but not sex; nor was there a significant sex x treatment interaction), suggesting the treatment was partially effective in increasing students’ self-efficacy beliefs. Seven of the twelve ANCOVAs indicated a statistically significant increase in the treatment group’s adjusted group mean self-efficacy belief scores as a result of being exposed to the intervention. In six of these seven analyses, increases in self-efficacy beliefs occurred in tasks that required three or more higher-order levels of thinking/learning. The results are discussed in terms of theoretical, empirical and practical significance. Future research is recommended to extend the intervention beyond the narrow confines of a Title I magnet school to settings where the intervention could be tested longitudinally, e. g., honors and gifted students, elementary and middle schools.

Workplace aggression: A reconceptualization of the construct & an exploration of strain based outcomes

The examination of Workplace Aggression as a global construct conceptualization has gained considerable attention over the past few years as organizations work to better understand and address the occurrence and consequences of this challenging construct. The purpose of this dissertation is to build on previous efforts to validate the appropriateness and usefulness of a global conceptualization of the workplace aggression construct. ^ This dissertation has been broken up into two parts: Part 1 utilized a Confirmatory Factor Analysis approach in order to assess the existence of workplace aggression as a global construct; Part 2 utilized a series of correlational analyses to examine the relationship between a selection of commonly experienced individual strain based outcomes and the global construct conceptualization assessed in Part 1. Participants were a diverse sample of 219 working individuals from Amazon’s Mechanical Turk participant pool. ^ Results of Part 1 did not show support for a one-factor global construct conceptualization of the workplace aggression construct. However, support was shown for a higher-order five-factor model of the construct, suggesting that it may be possible to conceptualize workplace aggression as an overarching construct that is made up of separate workplace aggression constructs. Results of Part 2 showed support for the relationships between an existing global construct workplace aggression conceptualization and a series of strain-based outcomes. Utilizing correlational analyses, additional post-hoc analyses showed that individual factors such as emotional intelligence and personality are related to the experience of workplace aggression. Further, utilizing moderated regression analysis, the results demonstrated that individuals experiencing high levels of workplace aggression reported higher job satisfaction when they felt strongly that the aggressive act was highly visible, and similarly, when they felt that there was a clear intent to cause harm. ^ Overall, the findings of this dissertation do support the need for a simplification of its current state of measurement. Future research should continue to examine workplace aggression in an effort to shed additional light on the structure and usefulness of this complex construct.^

Ensemble Stream Model for Data-Cleaning in Sensor Networks

Ensemble Stream Modeling and Data-cleaning are sensor information processing systems have different training and testing methods by which their goals are cross-validated. This research examines a mechanism, which seeks to extract novel patterns by generating ensembles from data. The main goal of label-less stream processing is to process the sensed events to eliminate the noises that are uncorrelated, and choose the most likely model without over fitting thus obtaining higher model confidence. Higher quality streams can be realized by combining many short streams into an ensemble which has the desired quality. The framework for the investigation is an existing data mining tool. First, to accommodate feature extraction such as a bush or natural forest-fire event we make an assumption of the burnt area (BA*), sensed ground truth as our target variable obtained from logs. Even though this is an obvious model choice the results are disappointing. The reasons for this are two: One, the histogram of fire activity is highly skewed. Two, the measured sensor parameters are highly correlated. Since using non descriptive features does not yield good results, we resort to temporal features. By doing so we carefully eliminate the averaging effects; the resulting histogram is more satisfactory and conceptual knowledge is learned from sensor streams. Second is the process of feature induction by cross-validating attributes with single or multi-target variables to minimize training error. We use F-measure score, which combines precision and accuracy to determine the false alarm rate of fire events. The multi-target data-cleaning trees use information purity of the target leaf-nodes to learn higher order features. A sensitive variance measure such as f-test is performed during each node’s split to select the best attribute. Ensemble stream model approach proved to improve when using complicated features with a simpler tree classifier. The ensemble framework for data-cleaning and the enhancements to quantify quality of fitness (30% spatial, 10% temporal, and 90% mobility reduction) of sensor led to the formation of streams for sensor-enabled applications. Which further motivates the novelty of stream quality labeling and its importance in solving vast amounts of real-time mobile streams generated today.