No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

Abstract A Keyword Spotting System (KWS) is a system that recognizes predefined keywords in spoken utterances or written documents. The objective is to obtain the highest possible keyword detection rate without increasing the number of false detections in a system. The common approach to keyword spotting is the use of a Hidden Markov Model (HMM). These are usually complex systems which require training speech data. The Typical HMM approach uses garbage templates or HMM models to match non-keyword speech and non-speech sounds. The purpose of this research is to design a simple Keyword Spotting System. The system will be designed to spot English words and should be easily adaptable to other languages There are many challenges in designing a keyword spotting system such as variations in speech like pitch, loudness, timbre that make recognition difficult. There can be wide variations in utterances even from the same speaker. In this research, the use of cross-correlation, as an alternative means for detecting keywords in an utterance, was investigated. This research also involves the modeling of a global keyword using a quantized dynamic time warping algorithm, which can function effectively with multi-speakers. The global keyword is an aggregation of the features from several occurrences of the same keyword. This research also investigates the effect of pitch normalization on keyword detection. The use of cross-correlation as a method for keyword spotting was investigated in both the time and MFCC domain. In the time domain the global keyword was cross-correlated with a pitch-normalized utterance. A zero lag ratio (the ratio of the power around the zero lag obtained from a cross correlation to the power in the rest of the signal is computed) was computed for each speech frame, a threshold was then used to determine if the keyword is present. For the MFCC domain the MFCC features of each keyword were computed, normalized and cross-correlated with the normalized MFCC features of portions of the utterance of the same size as the keyword. Cross-correlation of MFCC features of the keyword with that of each portion of the utterance yields a single value between 0-1. The portion with the highest value is usually the location of the keyword. Results in the time domain varied from keyword to keyword, some words showed a 60% hit rate while the average obtained from various keywords from the Call Home database had an average of 41%. Cross-correlation of the keywords and utterance in the MFCC domain yielded a 66% hit rate in test conducted on all different keywords in the Call Home and Switchboard corpus. The system accuracy is keyword dependent with some keywords having an 85% hit rate / Electrical and Computer Engineering

In January of 2012, Philadelphia Mayor Michael Nutter outlined the crime fighting measures that his administration would pursue during his second term as mayor. Included was a plan to introduce a multi-agency crime reduction program, which Philadelphia Police Commissioner Charles Ramsey and District Attorney Seth Williams would co-chair, called GunStat. GunStat was described as a collaborative effort to reduce gun violence through (1) identifying locations with a high incidence of violent crime, (2) pinpointing violent offenders responsible for these crimes, (3) focusing on arresting and prosecuting these offenders for crimes committed at these places, and (4) enhanced monitoring of offenders on probation and parole who are living and/or offending within these locations. In effect, GunStat was designed to target the right people (prolific, violent known offenders) at the right places (hot spots of violent crime). This dissertation is an in-depth, ex post facto evaluation of Philadelphia’s GunStat model as implemented over two phases and two years. It involved both a quasi-experimental research design which employed propensity score matching methods to generate comparisons, and a process-evaluation where several themes, including program implementation, were explored. The results here suggest that GunStat did not reduce crime relative to comparison locations. However, the qualitative data highlighted the importance of informal inter-agency networks that were developed during the course of the intervention, and suggested that GunStat put future collaborations on a solid footing. The implications for criminal justice policy, theory and evaluation design are discussed. / Criminal Justice

Background: Prostate cancer remains the second most common cause of cancer-related death in men in the United States (Siegel et al. 2017). Observational studies of patients with prostate cancer have found associations between diet and prostate cancer severity, but the evidence is inconsistent or inconclusive. The purpose of this thesis is to implement a validated international healthy diet score and evaluate whether or not it is associated with prostate cancer severity. Objective: The objectives of this thesis were: Chapter 1: examine whether an association exists between diet quality, using the validated Healthy Diet Score, and the severity of prostate cancer, and Chapter 2: examine the agreement between two methods of dietary data collection (an abridged FFQ and a longer previously validated FFQ) with respect to macronutrients and main food groups. Methods: We used observational data from the Randomized Intervention for Cardiovascular and Lifestyle Risk Factors in Prostate Cancer Patients (RADICAL PC), a multi-centre Canadian prospective cohort study into which men with a new diagnosis of prostate cancer or who were being treated with androgen deprivation therapy were enrolled. To complete objective 1 (Chapter 1) of this dissertation, a cross-sectional analysis was completed using baseline data collected in the RADICAL PC study. In order to evaluate the association of diet with prostate cancer severity, the relationship between the Healthy Diet Score and prostate cancer severity (stage and grade) was assessed. The second objective (Chapter 2) is a comparability sub-study comparing an abridged FFQ with a long, validated FFQ in a subgroup of participant (N=130) enrolled in the RADICAL PC study. Results: Chapter 1: In the cross-sectional analysis of baseline data collected in RADICAL PC, a higher diet score was not significantly associated with prostate cancer severity. An association between age and the high-risk prostate cancer category was found to be statistically significant (OR: 1.04, 95%CI 1.02-1.05, p<0.00). Chapter 2: There was good agreement between the abridged FFQ and long FFQ for carbohydrates, proteins, whole wheat, refined grains, fish, dairy, potatoes, fruits, nuts, and soft drinks (Spearman rank correlation >0.5). Food groups including fried foods, processed meats, vegetables and total fats (nutrients) were found to have moderate correlation (Spearman rank correlation between 0.3-0.5). There was low correlation for legumes, sugars and oils. Bland-Altman plots showed good absolute agreements between the two methods, and reliability test using Spearman’s correlation showed moderate to good correlation (0.45 to 0.75 among most food groups. Conclusion: There was no clear association between a healthier diet and prostate cancer severity in men with newly diagnosed prostate cancer. There was adequate agreement between the abridged SFFQ and the long FFQ of the expected food groups, and thus the SFFQ can be considered an appropriate tool to use for measuring diet among prostate cancer patients for some food groups and nutrients. / Thesis / Master of Health Sciences (MSc)

This work expanded on previous research on school effectiveness by developing and testing hypotheses about the specific relationships between school characteristics---including aggregated student and classroom characteristics---and student academic performance. The work used data from the "Early Childhood Transitions Project," a study of intensive social and educational services in a suburban school system, to identify and test the effect of a limited set of school-level characteristics on test score gains made by individual students on the Metropolitan Achievement Test (MAT) between the second and third grade. The analyses found that there are differences in the size of schools, the percent of low performing students, and the percent of students who are non-English speaking across the schools in the sample. Test score gains are affected by concentrations of these types of students at the schools. Students at schools in this sample with high concentrations of non-English speaking students or high concentrations of Hispanic students achieve lower test score gains than students in other schools. Another "concentration effect" emerged from the analysis of high-performing students in the sample. In particular, female students with high scores on the second grade MAT who are in schools with large concentrations of students who perform poorly on the second grade exam have smaller third grade test score gains than similar students who are in schools without a concentration of low performing students. These results suggest that more attention be paid to the influence that the characteristics of the student population have on the school's ability to implement the curriculum. As a first step, researchers may want to simply document the differences in the educational characteristics of students entering schools. This would provide evidence of the segregation that occurs across schools. Researchers may then want to conceptualize students within schools in terms of their homogeneity on demographic measures and their homogeneity on educational characteristics. This "educational minority or majority" concept may bring researchers closer to understanding the school environment, as it is organized by schools and experienced by children. / Ph. D.