16484 research outputs found
Sort by
Learning from Human Demonstration
No full textLearning from demonstration (LfD) is a branch of machine learning that focuses on learning how to perform a given task by observing a demonstrator perform one or several demonstrations of it. Moreover, many of the current LfD techniques assume a large pool of training examples from which to learn.The long term goal of our research is to develop general LfD methods which can more easily learn from human demonstrators than state-of-the-art methods, through requiring less training data and more carefully selecting when to obtain more training data. The main contribution of this thesis is a novel Active Learning from Demonstration algorithm called SALT (Selective Active Learning from Traces), an algorithm which can match or outperform other state-of-the-art algorithms in multiple domains, while using less training data. We have also gathered evidence that human demonstrators find it preferable to another state-of-the-art LfD algorithm, that it is less mentally burdensome for them to train, and that it learns better from them.Ph.D., Computer Science -- Drexel University, 201
Targeting Therapists' Experiential Avoidance During Exposure Delivery: An Experimental Investigation to Improve the Quality of Exposure-based Interventions
No full textDespite the large body of evidence that supports the efficacy of exposure therapy (ET) for anxiety disorders, many clinicians do not utilize exposure for these conditions, or they deliver exposure in an overly cautious, suboptimal manner. Research points to several therapist characteristics that might interfere with the decision to use exposure and with its optimal delivery: a persistent set of beliefs that ET is harmful and intolerable, experiential avoidance (EA) of anxiety and discomfort that arises during exposure, anxiety sensitivity, and intolerance of uncertainty. It is critically important to address these barriers to dissemination and implementation of ET. Though researchers have found promising results of didactic ET training that targets clinicians’ negative beliefs about exposure, no research has directly targeted therapists’ EA during didactic ET training. This study sought to determine whether incorporating techniques to improve therapists’ exposure-related willingness to experience anxiety into didactic training might improve the quality of their exposure delivery. Mental health clinicians and psychology graduate students attended a standard ET training workshop (ST; n = 53) that also targeted therapists’ concerns about ET or the same workshop with acceptance-based techniques (ST+ABT; n = 46) to improve therapists’ EA. Primary outcomes were assessed via the quality of participants’ exposure delivery as determined by a simulated exposure session. An additional dose of intervention was delivered at the conclusion at the exposure session via individualized feedback to participants about their exposure delivery. Secondary outcomes, including exposure delivery on a hypothetical case vignette, experiential avoidance, beliefs about exposure therapy, intolerance of uncertainty, and anxiety sensitivity were assessed after the workshop and at one-month follow-up. Reported frequency of exposure delivery after the workshop was also examined at the follow-up assessment. Results indicated that both groups delivered exposure therapy of equal quality according to objective ratings of therapists’ behaviors, but participants in the ST condition delivered better quality exposure according to a global subjective rating (p = .045). In contrast, there were no group differences in exposure delivery on the case vignette or actual use of exposure methods in the month after the workshop. Experiential avoidance scores did not improve from pre-workshop to post-workshop, though they improved by the follow-up assessment, and limited evidence suggests they improved to a greater degree in the ST+ABT group. Some evidence suggested that EA was associated with exposure delivery on the case vignette. In addition, negative beliefs about ET and intolerance of uncertainty were consistently associated with more cautious exposure delivery on the case vignette. No therapist factors were related to exposure delivery during the simulated session. Results suggest that the ST and ST+ABT protocols may have been more similar than distinct, and the ST+ABT protocol may have attempted to cover too much content in a half-day workshop; this may have resulted in greater consolidation of learning in the ST group. Results also suggest that therapists’ negative beliefs about ET and intolerance of uncertainty, and to some extent EA, may be critical barriers to effective exposure delivery. Future research should expand on efforts to identify which therapist factors most interfere with exposure delivery. Additionally, future research should examine the best methods for assessing therapists’ ET proficiency in real-world settings. Implications for the dissemination and implementation of ET are discussed.Ph.D., Psychology -- Drexel University, 201
Transparency or Over-Regulation: The FCC and its Role in Governing Public Access to Political Advertising Data
No full textOn April 23, 2012, the Federal Communication Commission (FCC) established a mandate that changed how broadcast stations were required to file their Public Inspection and Political Files. Broadcast stations have been very careful regarding who is granted access to their advertising financial records, and other station documents, as it is their main source of revenue.While this information has always been available in paper form, it is now conveniently available via the FCC’s website. This thesis analyzes the evolution of the FCC’s regulation of broadcast stations, regarding its level of increased oversight. The researcher used in-depth interviews with sales personnel and legal counselors from a top performing broadcast station in the Raleigh-Durham-Fayetteville television market.M.S., Television Management -- Drexel University, 201
Sampling algorithms for big graph analytics
No full textThe analysis of large graphs offers new insights into social and other networks, and thus is of increasing interest to marketeers, sociologists, mathematicians and computer scientists. However, the extremely large size of most graphs of interest renders them difficult to analyze because of at least four challenges: lack of memory, restricted access to the full graph, prohibitive computational cost and real-time changes in the graph. This dissertation presents graph sampling as a powerful and attractive approach to meet the above challenges, whereby properties of the full graph are estimated based on an examination of only a small portion of the graph. In this dissertation, we focus on two graph sampling strategies: edge-based sampling and traversal-based sampling. For edge-based sampling, we propose an edge-based sampling framework for big-graph analytics in dynamic graphs. It enhances the traditional model by enabling the use of additional related information. To demonstrate the advantages of our proposed framework, we present a new sampling algorithm which provides an unbiased estimate of the total number of triangles in a fully dynamic graph where both edge additions and deletions are considered. Our algorithm addresses three of the aforementioned challenges; it has low memory and computational costs, and can be applied to dynamic graphs. In particular, it offers a significantly improved performance in real time estimates compared to current state-of-the-art methods. We also propose several traversal-based graph sampling algorithms for the estimation of a micro-structural property (motif statistics) and the estimation of a macro-structural property (the two largest eigenvalues of the graph). All of these algorithms solve the challenges of prohibitive computational and storage costs, and restricted access. For micro-structural property, we develop a new sampling algorithm which estimates the concentration of mo- tifs of any size via random walk. Unlike previous approaches which enumerate subgraphs around the random walk to find motifs, our algorithm achieves its computational efficiency by using a randomized protocol to sample subgraphs in the neighborhood of the nodes visited by the walk. The experimental results show that our algorithm achieves better accuracy and higher precision than previously known algorithms. For macro-structural property, we propose a series of new sampling algorithms which estimate the top eigenvalues of a graph. Unlike previous methods which try to collect a subgraph with the most influential nodes, our algorithms achieve estimates of the two largest eigenvalues by estimating the number of closed walks of a certain length. The experimental results show that our algorithms are much faster and achieve higher accuracy on most graphs than previously known algorithms seeking to address the same challenges.Ph.D., Electrical Engineering -- Drexel University, 201
Dare or dare not: The impact of power pose on consumers' price perceptions
No full textThis dissertation research examines how an advertising model’s pose that signals power (hereafter referred to as power pose) influences consumers’ price perceptions. Prior research on social judgments suggests that people generally indicate submissive behavior (e.g., averting gaze and look down) when encountering with a powerful subject. Yet, little attention has been paid to how this affects consumers when marketing information is being processed. Across six studies, this research finds that displays of a model’s non-verbal, dominant gestures stimulate consumers to selectively attend to and process price information. Specifically, when a model exhibits a high-power pose (vs. low-power pose), consumers are more likely to pay attention to and process price information of the products that are displayed on the bottom of an advertisement. As such, consumers indicate better price recall for the outfits and accessories displayed on the lower body. Consumers also more accurately estimate the price discount depth when the comparative price promotion appears below the image of the model in a high-power pose. In contrast, consumers indicate better price recall ability and estimation of the discount when they are visualized above the model in a low-power pose. This effect of power pose occurs both when an image of a model is presented and when an actual model displays power, whether high or low, through their pose. However, the effect of power pose reverses when the model’s face is not visible. When the face is eliminated, consumers are not dominated by the power of the pose, and as such do not display submissive behavior. Instead they attend more to the model’s upper body than the lower body to garner more information about the model. Therefore, consumers indicate better performance in recalling price information of apparel displayed on the upper body as opposed to those on the lower body when they encounter a high-power pose with the absence of the face. The mediating role of anxiety is introduced as a potential underlying mechanism. This research provides insights to help marketers identify ideal locations for displaying price information when using a model in advertising, and for in-store displays.Ph.D., Marketing -- Drexel University, 201
Predictors of Hemoglobin A1c Among Adults 40 to 59 Years of Age in the United States: National Health and Nutrition Examination Survey 2003 to 2004 and 2013 to 2014
No full textObjective: The risk for type 2 DM increases with age, becoming more significant after 45 years of age. The present research is intended to expand upon the current literature by investigating whether dietary magnesium intake, indirect measures of body composition (Body Mass inDex (BMI), waist circumference, waist-to-height ratio) and sedentary behavior activity are predictive of hemoglobin A1c percentage. Research Design and Methods: We used cross-sectional data, including adults 40 to 59 years of age, from the National Health and Nutrition Examination Survey (NHANES) for 2003 to 2004 and 2013 to 2014. Responses to survey questions for dietary intake, anthropometric measurements (BMI, waist circumference, and waist-to-height ratio), and responses to the Physical Activity Questionnaire regarding time watching television or videos, and time spent sitting in front of a computer per day were analyzed. A binary logistic regression analysis was used to determine whether dietary magnesium, BMI, wait circumference, waist-to-height ratio or sedentary behavior time were predictors of a HbA1c ≥ 6.5%. The regression model was further adjusted for age, sex, race and ethnicity. Results: Using a univariate binary logistic regression analyses, the odds of having a HbA1c measure of ≥ 6.5% was not associated with mean dietary magnesium intake (NHANES 2003 to 2004 [p=0.873]; NHANES 2013 to 2014 [p=0.534]). Adding age, sex, race and ethnicity into the model, the odds of having a HbA1c of ≥ 6.5% were not associated with mean dietary magnesium intake (NHANES 2003 to 2004 [p=0.310]; NHANES 2013 to 2014, [p=0.530]). Measures of BMI, waist circumference and waist-to-height ratio were lower between NHANES 2003 to 2004 and NHANES 2013 to 2014 (p=0.0001). The odds of having a HbA1c measure of ≥ 6.5% was significantly associated with having a greater BMI, waist circumference and waist-to-height ratio (p=0.0001). In the 2003 to 2004 NHANES sample multivariate model only, waist-to-height ratio was associated with greater odds of having a measured HbA1c of ≥ 6.5% (OR: 2.91, 95% CI: 1.69, 5.04). In a univariate model, adults reporting ≥ 8 hours of SBT in NHANES 2003 to 2004 had 2.02 increased odds of a HbA1c ≥ 6.5% (OR= 2.02, 95% CI: 1.31, 3.13, p<0.0001) compared to adults reporting ≤ 3 hours. After adjusting the regression model for age, sex, race, ethnicity and body mass index, adults reporting ≥ 8 hours of SBT in NHANES 2003 to 2004 had 1.72 increased odds of HbA1c ≥ 6.5% (OR= 2.02, 95% CI: 1.10, 2.68, p<0.0001) compared to adults reporting ≤ 3 hours of SBT. Reported SBT was not a predictor of HbA1c ≥ 6.5% for NHANES 2013 to 2014. Conclusion: Dietary magnesium was not a predictor of HbA1c ≥ 6.5% among adults, 40 to 59 years of age, in NHANES 2003 to 2004 and 2013 to 2014. Waist circumference and waist-to-height ratio were predictors of HbA1c ≥ 6.5% among adults, 40 to 59 years of age, in NHANES 2003 to 2004 but not in NHANES 2013 to 2014.SBT was a predictor of HbA1c ≥ 6.5% among adults, 40 to 59 years of age, in NHANES 2003 to 2004, but was not a predictor in NHANES 2013 to 2014.Ph.D., Nutrition Sciences -- Drexel University, 201
Humane Slaughter Does Not Exist: How Regulatory Policies Have Failed to Protect Animals in Slaughterhouses
No full textThe Humane Methods of Slaughter Act (HMSA) was created to protect animals from egregious forms of cruelty in United States slaughterhouses. Under HMSA means of slaughter must be rapid, effective, and performed before the animal is processed. In addition, the animal must be rendered insensible to pain. It is important to note who is protected under HMSA, as the act only applies to the following animals: cattle, calves, horses, mules, sheep and swine. ' Unfortunately, various outlets including Government Accountability Office, Office of the Inspector General, and first person accounts, have proven HMSA is not achieving its intended goal of protecting animals from egregious acts of cruelty. This case study makes recommendations to improve policy surrounding the treatment of animals in slaughterhouses and cites specific examples as evidence for these policy recommendations. Most significantly, this study seeks to understand how regulatory policy has failed in reducing harm to animals prior to slaughter.M.S., Public Policy -- Drexel University, 201
Moving Through Viewpoints in Neurorehabilitation: The Development of a Dance/Movement Therapy Clinical Method for Individuals with Traumatic Brain Injury and Stroke
No full textA dance/movement therapy (DMT) clinical method is proposed to address treatment goals and objectives of individuals suffering from traumatic brain injury (TBI). There is a need for more clearly outlined DMT interventions with the TBI population in the published literature. Viewpoints is an acting and choreography approach that could be a useful tool in DMT. It provides an organized structure for interventions that could address specific deficits of brain injury within the cognitive, physical, and psychosocial sequelae. Patients suffering from a TBI or stroke often lose cognitive awareness and sense-perception. These individuals must be reoriented to time and space in order for the brain to heal and allow for a successful recovery in the rehabilitation process. Viewpoints expands awareness and ways of moving and relating to the self, others, and the environment through the exploration of time and space elements. Clinical vignettes with TBI and stroke patients are included to illustrate technical choices of Viewpoints structure and process within the DMT session. Viewpoints in DMT attempts to enhance body and spatial awareness, and expand movement repertoire as well as creative decision-making and expression. Recommendations for future clinical practice, research and adaptations for other therapeutic contexts are considered.M.A., Dance/Movement Therapy and Counseling -- Drexel University, 201
Automatically Measuring Software Architecture and Identifying Architecture Problems
No full textSoftware architecture is critical to a software system. As software evolves, complexity accumulates through various maintenance activities: such as bug fixes, feature additions, etc., inevitably resulting in architecture degradation that negatively impacts a system’s maintainability. Despite decades of research about software measurement and analysis, it is still a challenge for development teams to reliably measure the maintainability of software architecture and precisely diagnose architecture problems incurring maintenance difficulties. Numerous software metrics have been proposed to measure software quality, but they haven’t demonstrated the reliability to indicate architecture maintainability, nor to compare the maintainability of different projects. Bug prediction, code smells and antipatterns have been proposed to predict defective files and detect design flaws. However, bug prediction studies never considering architecture problems among the detected files, which are the root causes of maintenance pain; code smells detection techniques tend to report a large number of problems, many of which are false positives; detecting anti-pattern heavily depends on the skill of architecture analysts. These make the existing techniques hard to be used to precisely find the true problems causing maintenance difficulties. This dissertation presents our methodology that advances our ability to monitor the variations of software architecture maintainability by using reliable measurement, and identify the architecture flaws that should be addressed to reduce maintenance difficulties. Our proposed methodology consists of three parts as follows: 1) we proposed a novel metric which has presented similar properties as real-word metrics, and could be used to reliably measure software maintainability. Based on the proposed metric, we created an industrial benchmark serving as a “health chart” of software maintainability. Like real-world growth chart, managers and architects could compare their projects with the benchmark to assess the maintainability level of their projects, the abnormal results would signal a early symptom of architecture degradation; 2) We also proposed a history measure, which indicates how well the maintenance tasks could be implemented independently during evolution process. This history-based measure could help to monitor the real-time interactions of maintenance tasks, which can not be reflected by syntax-based metrics; 3) Software measurements provide us a coarse assessment of the maintainability of software architecture. To pinpoint the root causes of maintenance difficulty, we then automatically identified a suite of architecture design flaws, the real architecture problems which have caused high maintenance costs in software system. Our studies have shown that our methodology could faithfully measure maintainability of software architecture and precisely diagnose the true architecture problems in a project, which helps development teams decide if, when, where and how to refactor.Ph.D., Computer Science -- Drexel University, 201
The Nature, Severity, and Treatment of Insomnia Disorder in Relapsing-Remitting Multiple Sclerosis
No full textObjective. Examination and treatment of insomnia in those with multiple sclerosis (MS) is crucial, as insomnia may contribute to decreased quality of life and disease progression through increased inflammatory responses; however, few studies have adequately assessed for insomnia among individuals with MS. Similarly, few studies have examined the correlates of insomnia or implemented behavioral treatments for insomnia. This study aimed to characterize the nature of insomnia, to explore the symptom profiles of those with insomnia, and to conduct a pilot trial of Brief Behavioral Treatment for Insomnia (BBTI) among those with insomnia disorder and relapsing-remitting multiple sclerosis. Procedures. 31 participants attended an initial in-person session where they completed the Duke Structured Interview for Sleep Disorders and several self-report measures to characterize sleep, health, and mood. Subsequently, participants completed two weeks of sleep diaries. Four participants diagnosed with insomnia disorder during the initial session completed a trial of BBTI. Results. Findings indicated that 54.8% of the study sample met criteria for insomnia disorder. Insomnia severity was related to pain and physical health-related quality of life. Insomnia severity was also related to fatigue; however, when sleep aid use was controlled for, the association between insomnia and fatigue was no longer significant. Lastly, three individuals had clinically significant improvements in insomnia and decreased fatigue after BBTI. Discussion: Insomnia is prevalent among those with RRMS and is related to factors of MS that are known to be debilitating. BBTI may serve as an effective treatment to improve both insomnia and fatigue; however, larger studies to demonstrate efficacy are necessary.Ph.D., Psychology -- Drexel University, 201