12466 research outputs found
Sort by
Optimization of the Placement of the Ultrasound Scanlines on the Forearm for an Upper Limb Prosthesis
Get PDFSonomyography is an emerging technique that uses ultrasound to detect muscle deformation and is being explored as a real-time alternative to surface electromyography for deriving control signals from functional activity. Many groups have demonstrated the feasibility of using commercial ultrasound systems to control upper limb prostheses; however, these systems are bulky and not optimized for wearable use. In this study, a novel 4-channel ultrasound system with miniaturized electronics optimized for forearm applications has been used. While previous work has demonstrated that data from 4 channels may be sufficient to classify multiple grasps, the performance may be dependent on the anatomical placement of the individual transducers on the forearm. In this study, we evaluated the effects of transducer placement on classification performance and explored different measurements to determine optimal anatomical region for placement. These metrics consisted of Mutual Information (MI), Structural Similarity Index (SSIM), and Sum of Squared Distance (SSD); which quantify the amount of information is derived from every ultrasound transducer. Ultrasound M-mode images of different hand/wrist gestures were collected from 4 subjects with ultrasound transducers placed at three different positions on the forearm. The first position was at the flexor muscles, the second observing the extensor muscles, and the third was a custom placement of the transducers targeting specific muscle compartments and regions on the forearm. MI/SSIM/SSD were used to calculate how much information each ultrasound transducer contained, and the values were correlated to the performance of a LDA classifier's ability to differentiate between the gestures. The results show that the LDA was able to discriminate between the different hand gestures with an average accuracy of 76.4 ± 4.04% for the extensor muscle position,
97.5 ± 1.72% and 99.4 ± 0.61% for the flexor muscles and custom targeted position, respectively. No correlation was found between MI and the classification performance. Strong statistically significant correlation was found between SSD and SSIM values and classification performance (p-value < 0.001) . This study demonstrates the feasibility of using 4-channel single element M-mode ultrasound transducers to recognize complex hand gestures and emphasizes the importance of targeting specific muscle compartments and regions on the forearm to obtain high classification accuracy
Learning Semantic Representations and Visual Navigation in Indoor Scenes
Get PDFIn the past decade, computer vision has made remarkable strides in extracting semantic and geometric information from images. These advancements were driven by deep learning techniques and large datasets and led to the development of effective object detection and scene semantic parsing approaches. Recent shift towards Embodied Artificial Intelligence focuses on the application of existing semantic representations within real-world agents (e.g., a robot), accomplishing tasks such as navigation or object search. In this thesis, we demonstrate that integrating these semantic representations with the agent's decision-making capabilities can significantly enhance the sample efficiency and improve the generalization of end-to-end navigation models. This thesis makes three primary contributions.First, we study the problem of visual servoing in the reinforcement learning framework and introduce a trainable end-to-end visual servoing model for the target object and image-goal navigation tasks. Second, we present a novel approach to detect unknown out-of-distribution objects not covered in the training data by leveraging pixel-level predictions obtained by semantic segmentation models. Finally, we consider the problem of time-limited robotic exploration in previously unseen environments where exploration is limited by a predefined amount of time. We propose a novel exploration approach using learning-augmented model-based planning where we exploit semantic mapping to estimate frontier properties
Investigation of the Correlation between Screen Time, Social Media Status, and BMI Status among Mason College Students
Get PDFPeople's lives have become increasingly reliant on technology, especially with the emergence of social media. Research has shown that high technology usage has a detrimental impact on health and is linked to rising rates of overweight individuals and obesity worldwide (Chau et al., 2014; Liu et al., 2021; Melton et al., 2014). There is a surge in the number of college students who use technology. Previous research examined the effect of technology on BMI in children and adolescents (Alotaibi et al., 2020; Rosen et al., 2014; Shen et al., 2021), but to date, there has been limited research conducted on young adults ages between 18 -24. Increased technology usage is one of the main culprits that lead to poor dietary choices and sedentary lifestyles, both of which have been linked to an increase in BMI (Chau et al., 2014; Liu et al., 2021; Melton et al., 2014). This study aims to examine the relationship between screen time usage and an increase in body mass index (BMI) among George Mason University college students. The data was abstracted from the Health Starts Here Study and included 131 first year students from George Mason University. Technology usage, BMI, diet, physical activity, and sleep were obtained through various questionnaires and anthropometric measurements. These variables were analyzed using different statistical tests: Person's correlation, independent t-tests, and stepwise regression analysis. The results showed that there is a non-significant correlation between using social media, BMI, and dietary choices. Person's correlation result revealed that the correlation between social media and BMI was r (129) =.072 p=.416 while the correlation between BMI and diet was r (129) =.09 p=.30. T-test result showed on average low-tech usages (M#.24, SD=5.09) had lower BMI scores than high-tech usage (M$.71, SD=6.81). This study concludes that the amount of time of using social media has no impact on increasing BMI
An Examination of Affect-Related Brain Activity and Substance Use Among Adolescents
Get PDFDeath and disability related to substance use disorder have increased substantially over the past couple of decades. Most adults with substance use disorder initiated substance use as adolescents, making adolescence a critical period for the prevention of substance use and substance use disorder. It is therefore important to identify risk factors for adolescent substance use. Recent research has demonstrated the role of altered affective processing in adolescent substance use. Unfortunately, most of this research has employed self-report and behavioral methods, which, while valuable, are limited in comparison to other methods, namely functional neuroimaging, in detecting subtle neural-level differences in affective processing and how it relates to adolescent substance use. Thus, the focus of this dissertation is on neural affective processing and adolescent substance use employing functional neuroimaging. In Study 1 of this dissertation, a systematic review of neuroimaging studies examining affective processing and adolescent substance use was conducted. Results revealed that higher activation in midcingulo-insular regions—particularly the striatum—to positive affective stimuli (e.g., monetary reward) was most often associated with initiation and low-level use of substances, whereas lower activation in these regions was most often associated with substance use disorder and higher-risk substance use. In regard to negative affective stimuli, most research demonstrated associations between higher activation of midcingulo-insular network regions and adolescent substance use. Associations between activation in additional network regions (e.g., frontoparietal, pericentral) and adolescent substance use were mixed. To extend findings from Study 1, Study 2 of this dissertation was an empirical study examining how patterns of neural activation in two standardized and one naturalistic affective processing tasks classify substance use as well as predict substance use intentions and expectancies in 11–15-year-old adolescents (n = 168). Machine learning analyses were performed. Results did not provide evidence that neural activation to negative or positive affective stimuli—neither standardized nor naturalistic—could reliably classify adolescent substance use and predict adolescent substance use intentions and expectancies. Implications of all findings, as well as limitations and directions for future research are discussed
Evaluating New Methods to Detect Threatened and Cryptic Manatee Populations
Get PDFThe conservation status of approximately 21% of marine mammals has not yet been assessed, despite the vast number of anthropogenic threats they face at local, regional, and global scales, with the primary threats being bycatch in fisheries, pollution, and illegal hunting. Several factors make detecting marine mammals exceptionally difficult in comparison to terrestrial mammals, including cryptic behavior, such as long dive times and poor water clarity. Furthermore, anthropogenic threats can exacerbate cryptic behavior of marine mammals. Manatees (Trichechus spp.) are examples of marine mammals that are considerably difficult to detect, especially where these species are still illegally hunted, due to their cryptic nature and ability to inhabit heterogenous aquatic environments. Currently, aerial surveys are the predominate method for monitoring manatee populations (as well as other marine mammal populations such as cetaceans). However, due to safety and logistical concerns, many manatee populations are not able to be routinely surveyed as is the case for African manatees (T. senegalensis). As a result, existing methods to detect marine mammals may not be sufficient to accurately survey populations to obtain information critical to establishing effective conservation efforts. The main goal of this dissertation is to evaluate two new methods that stem from advancements in technology to detect cryptic manatee populations: 1) the use of unmanned aerial vehicles (UAVs) and 2) environmental DNA (eDNA) from residual saliva remaining on foraged upon aquatic vegetation. To do so, I first performed a comparative analysis of existing methods (indirect observations as proxies for manatee presence and boat-based visual observations) with the use of a small, multi-rotor UAV to detect African manatees in two lagoon waterbodies in southwest Nigeria where the species is still heavily hunted. I measured environmental variables hypothesized to affect manatee presence at a survey site and associated these variables with detection methods. Finally, in this study, I also used the UAV to monitor real-time threats to the species at the survey sites. The results of this study showed that the indirect observation method results in the greatest number of detections, but UAV surveys resulted in twice the number of detections in comparison to boat-based visual observations. Water depth was the most influential variable associated with detecting African manatees using a UAV and with indirect observations, although the direction of the effect differed between the two methods. Using the UAV, the most abundant threats to African manatees that were observed included the presence of fishermen and fishing nets. In the second study, I further evaluated survey effort associated with conducting UAV surveys to detect Amazonian manatees (T. inunguis) in a controlled environment at a closed man-made lake being used to rehabilitate a known number of individuals and evaluated the effect of various environmental variables on detection probability. In this study, I was able to calculate the number of repeat surveys (n = 3) to be confident that true abundance was established using a UAV. I also found that environmental variables influencing behavior of individuals (i.e., water temperature) had the greatest effect on detection probability in comparison to habitat use or environmental variables associated with visibility within the water column. This is the first time a UAV has been used to detect either species. Finally, the third study conducted aimed to determine the efficacy of using residual saliva remaining on foraged upon aquatic vegetation, a new source of eDNA for aquatic species. I collected ex-situ eDNA samples by hand-feeding captive Florida manatees (T. m. latirostris) kale and from collecting recently foraged upon water lettuce (Pistia stratiotes) from the captive Amazonian manatees in the second study. I also collected eDNA samples from water hyacinth (Eichhornia crassipes) and various grasses (Brachiaria spp.) hypothesized to have been foraged upon by wild African manatees at the study sites in Nigeria from the first study. I successfully detected manatee eDNA for the first time from vegetation samples across all study sites. Overall, this dissertation resulted in protocols developed for two new survey methods to detect cryptic manatee species in difficult-to-detect environments. Continued evaluation of these methods has the potential to develop range-wide surveys to better inform management of understudied and threatened aquatic species, such as manatees
Towards Robust and Privacy-aware Time Series Data Mining
Get PDFWith the advancement of sensor technology, large volumes of high-resolution time series are collected in a variety of domains and data mining in large-scale time series has attracted great research attention. While many time series data mining methods in large-scale time series are able to achieve satisfactory performance in a wide variety of tasks in ideal cases, they suffer from performance decreases or even complete failure because of environmental noise, unexpected signals, and potential system failures in many more complex real-world applications. In addition, there have been rising concerns about privacy issues for performing data mining since it is often required to have data access to perform certain data mining tasks on high-resolution time series. However, numerous research efforts have found that long shape-based patterns embedded high-resolution time series could contain sensitive information and be misused by a malicious modeler. However, despite there is a large body of privacy literature, how to perform time series data mining tasks while protecting sensitive patterns is surprisingly seldom explored in privacy-preserving literature. In this dissertation, I investigate the robustness issues in two popular time series data mining tasks, time series anomaly detection and time series chain discovery. To mitigate the challenges in robustness, I introduce three new robust methods: a new time series anomaly detection method that resists unknown background patterns; a new chain discovery method that works for time series with a gap or a corrupted segment; a more general chain discovery method that is more robust to noise and pattern fluctuation in more complex time series. To address the privacy issue, I take the first step to investigate the pattern-level privacy problem in time series data mining tasks. I introduced a privacy-aware scheme that allows time series data to be shared between data owners and a service provider via an intermediary data structure without leaking sensitive information about the owners. My proposed method protects sensitive patterns and their locations and supports various downstream time series data mining tasks while maintaining the performance of these tasks
It Began as Tides on the Magothy
No full textThis thesis has been embargoed for 10 years. It will not be available until April 2032 at the earliest.This thesis is the beginning foundation of a larger work that focuses on my life as a young child, navigating being abandoned by my mother, Janet Carroll Smith, at an early age, and handing the responsibilities the caring for my brothers and working through my own trauma as a teenager. This is the first section of a three-section project that centers on the activity and dysfunctional life that was thrown on three young children. This is a story of heartbreak, fortitude, and love that explores the dynamics between mother and father, father and daughter, and mother and daughter. The beginning section of this thesis is part a larger work that will further include my mother’s best left through her late 40’s and 50’s, to being diagnosed with dementia, where my family and I would take on the responsibility and care of her until her passing in 2015.2032-04-1
Machine Learning Automation for Virtual Reality
No full textVirtual Reality (VR) game development techniques are relatively new in relation to conventional 2-dimensional (2D) content. Although there has been significant research conducted in this new field, more work is still needed as there are still some prevalent issues. A significant issue reported by some users is that the perceived difficulty of a game can vary drastically between users. This is because the nature of VR gives more autonomy to users and lets them play games differently than the developer might've intended. To address this, I have proposed a system that tracks user difficulty perception on the manipulation of various game parameters that affect difficulty. The collected user data is used to train a machine learning regressor to predict the perceived difficulty of different game levels. The initial findings show a 53% prediction error. However, further analysis has shown that the predictions are realistic and adequate. Anomalies in prediction are explainable and prediction error can be reduced to 26% through the removal of some outliers. Limitations of this work, like the limited dataset size, are also addressed for future work to improve accuracy and performance. This thesis was primarily written with future work in mind, as the addressed problem is complex and requires further examination for a final and applicable model. The final model proposed uses MCMC optimization and is aimed at automating optimization of game parameters to tailor experiences to intended difficulty and/or emotions. Thus, the main contribution of this paper is its address of an insufficiently covered issue by producing a key approach and proposing detailed suggestions for future research
Reflection
Get PDFMy thesis work focuses on gender identity, religion, and speculative design. The thesis exhibition and body of work is simply named REFLECTION for its quality to reflect the identity of myself and the identities of others. This work began four years ago when investigating topics that many viewers harbor misconceptions about: religious faith, transgender, and in the unexpected places where they meet. In response, I have designed and built hypothetical religious objects that both describe and offer solutions to the various questions and struggles I face as a trans man, Lutheran, and human being. The work I created examines the possibilities of being transgender in a profound and endearing way
The Experiences of Prostitution Court Participants and Social Bond Theory
Get PDFProstitution Diversion Programs are amongst the least prevalent and under-studied types of problem-solving courts in the United States. This has subsequently led to a knowledge gap in the understanding of prostitution court participants, how they perceive the program, and what factors participants believe aid in their success. This study analyzes interview transcripts from seven prostitution court participants in two courts within Baltimore and Philadelphia. I first identify how the components of Social Bond Theory (SBT) appear in the lives of these participants and then use these components to explain how prostitution court participants interpret their experiences in treatment. The components of attachment, commitment, and involvement are discussed in the findings and the theoretical/practical implications are acknowledged