93887 research outputs found
Sort by
A Critical Examination of Spatial Skills Assessment: Validity, Bias, and Technology
No full textAt the highest level, this dissertation is a case study on how bias can become encoded into the tools used to measure a construct and into the very definition of the construct itself. In this case, the construct is spatial ability. This dissertation focuses on the validity and accuracy of spatial tests and illuminates gender bias that is interwoven with the history of spatial testing.First, I present a critical analysis of the graphical imagery used in spatial tests and explain why the imagery may be unclear and lead the tests to be inaccurate. I analyzed a collection of research in which researchers modified the stimuli used in spatial tests and found that the tests became easier when the imagery was made clearer. Thus, I conclude that imagery presentation impacts test difficulty, a likely example of construct-irrelevant variance which may reduce the validity of some spatial skills assessments and introduce bias in favor of individuals with past experience in engineering graphics, who historically are more likely to be men.Second, I make a critical review of gender differential research in spatial skills. I argue that the construct of “spatial ability” has been co-constructed with gender, in that it has been devised in a manner influenced by gender beliefs. Because of a preexisting belief that men had better spatial skills than women, some test creators “selectively bred” spatial instruments to produce the expected gender differences. Such instruments, including the very popular Mental Rotation Test (MRT), cannot validly assess between-group differences. Biological or evolutionary explanations for sex differences in spatial ability lack empirical evidence. Instead, the differences are rooted in the shaping of the construct of “spatial ability” to create the expected gender patterns.Finally, I describe an experiment designed to investigate the hypothesis that using a spatial test with content from a feminized discipline will show different patterns in gender differences. Female engineering students outperformed males on the Digital Apparel Spatial Visualization Test (DASVT), while the male engineering students scored higher than the female students on the Purdue Spatial Visualization Test (PSVT:R). Students with relevant background experience scored better than students without experience on both assessments. The results demonstrate the shortcomings of using a single instrument to assess a concept as heterogeneous as spatial skills. I conclude this dissertation with a discussion of the implications of my work and recommendations for researchers and educators
Determination of Lake Water Level Using Space Laser Altimetry
No full textThe spaceborne lidar Ice, Cloud, and land Elevation Satellite (ICESat)-2 provides the ATL13 data product for inland water bodies. However, its quality characteristics are not yet fully understood. This study presents a robust method for extracting lake water level data and makes a comprehensive evaluation on the determined water levels. The selected study areas are Lake Huron and Lake Superior, which are part of the Great Lakes. The extracted water levels from ATL13 over a period of four years are validated by using the field measurements at the closest NOAA hydrological stations. The evaluation is carried out in terms of data specifications, wind speed, frozen precipitation, distance of photon segments to hydrological stations, data acquisition time, and beam intensity. The determined water levels are then further used for seasonal monitoring and modeling of water surface. This work demonstrates the critical need for outlier removal and the capability of the ATL13 data. A total bias of 9 - 10 cm is found in the ATL13 product. It is found that frozen precipitation can lead to an overestimation (~ 5 cm) of the water level. However, the uncertainty of water level determination is not found to be significantly related to the laser beam intensity and data acquisition time. We expect that these findings will be valuable for users employing the ATL13 inland water body product and for developers producing future versions of the ATL13 product
Efficient and Consistent Convolutional Neural Networks for Computer Vision
No full textConvolutional Neural Networks (CNNs) are machine learning models that are commonly used for computer vision tasks like image classification and object detection. State-of-the-art CNNs achieve high accuracy by using many convolutional filters to extract features from the input images for correct predictions. This high accuracy is achieved at the cost of high computational intensity. Large, accurate CNNs typically require powerful Graphics Processing Units (GPUs) to train and deploy, while attempts at creating smaller, less computationallyintense CNNs lose accuracy. In fact, maintaining consistent accuracy is a challenge for even the state-of-the-art CNNs. This presents a problem: the vast energy expenditure demanded by CNN training raises concerns about environmental impact and sustainability, while the computational intensity of CNN inference makes it challenging for low-power devices (e.g. embedded, mobile, Internet-of-Things) to deploy the CNNs on their limited hardware. Further, when reliable network is limited or when extremely low latency is required, the cloud cannot be used to offload computing from the low-power device, forcing a need to research methods to deploy CNNs on the device itself: to improve energy efficiency and mitigate consistency and accuracy losses of CNNs.This dissertation investigates causes of CNN accuracy inconsistency and energy consumption. We further propose methods to improve both, enabling CNN deployment on low-power devices. Our methods do not require training to avoid the high energy costs associated with training.To address accuracy inconsistency, we first design a new metric to properly capture such behavior. We conduct a study of modern object detectors to find that they all exhibit inconsistent behavior. That is, when two images are similar, an object detector can sometimes produce completely different predictions. Malicious actors exploit this to cause CNNs to mispredict, while image distortions caused by camera equipment and natural phenomena can also cause mispredictions. Regardless the cause of the misprediction, we find that modern accuracy metrics do not capture this behavior, and we create a new consistency metric to measure the behavior. Finally, we demonstrate the use of image processing techniques to improve CNN consistency on modern object detection datasets.To improve CNN energy efficiency and reduce inference latency, we design the focused convolution operation. We observe that in a given image, many pixels are often irrelevant to the computer vision task – if the pixels are deleted, the CNN can still give the correct prediction. We design a method to use a depth mapping neural network to identify which pixels are irrelevant in modern computer vision datasets. Next, we design the focused convolution to automatically ignore any pixels marked irrelevant outside the Area of Interest (AoI). By replacing the standard convolutional operations in CNNs with our focused convolutions, we find that ignoring those irrelevant pixels can save up to 45% energy and inference latency.Finally, we improve the focused convolutions, allowing for (1) energy-efficient, automated AoI generation within the CNN itself and (2) improved memory alignment and better utilization of parallel processing hardware. The original focused convolution required AoI generation in advance, using a computationally-intense depth mapping method. Our AoI generation technique automatically filters the features from the early layers of a CNN using a threshold. The threshold is determined using an Accuracy vs Latency curve search method. The remaining layers will apply focused convolutions to the AoI to reduce energy use. This will allow focused convolutions to be deployed within any pretrained CNN for various observed use cases. No training is required
River Restoration Intelligence and Verification (RRIV): Development of a Low-Cost, Versatile Embedded System for Broad-Scale Monitoring of Water Quality and Greenhouse Gas Emissions
No full textSensor technology is evolving rapidly, offering new opportunities for environmental data collection. Yet, despite the large number of sensors now available, there is a lack of logging platforms that can be used to operate these sensors in situ. To address this shortfall, River Restoration Intelligence and Verification (RRIV) has developed an environmental data logger that meets the needs of the environmental sensing community. This platform has several advantages that reduce the time, effort, and technical know-how required to deploy environmental sensors. An extensive low-power mode is available, and hardware such as a real-time clock with an independent power source is incorporated. A driver system has been developed that allows users to incorporate sensors into the platform with minimal effort. RRIV loggers also include a command line interface that allows user to add or remove sensors, calibrate sensors, or configure deployments without the need for C/C++ programming, something that is not possible with outof-the-box microcontrollers such as Arduino and ST Nucleo products. The technology incorporated into RRIV and how it is applied and deployed in the field is described. This includes a description of power consumption. Protocols and descriptions of case construction are also included. RRIV loggers configured to monitor carbon dioxide and methane are used to demonstrate how this platform is used in the field
Heterogeneous Structural Elements Based on Mechanics of Structure Genome
No full textThe Mechanics of Structural Genome (MSG) is a unified homogenization theory used to find equivalent constitutive models for beam, plate, and solid structures. It has been proven accurate for periodic structures. However, for certain applications such as non-prismatic wind turbine blades and helicopter flexbeams featuring ply drop-off, where there is no repeating structure and the periodic boundary condition cannot be used, MSG’s accuracy is limited. In this work, we aim to extend MSG to find element stiffness matrices directly for aperiodic structures, instead of beam properties or three-dimensional (3D) solid material properties. Two finite elements based on MSG have been developed: Heterogeneous Beam Element (HBE) and Heterogeneous Solid Element (HSE).For beam modeling, the beam-like structure is homogenized into a series of 3-node Heterogeneous Beam Elements (HBE) with 18 × 18 effective beam element stiffness matrices. These matrices are used as input for one-dimensional (1D) beam analysis using the Abaqus User Element subroutine (UEL). Using the macroscopic beam analysis results as input, we can also perform dehomogenization to predict the stresses and strains in the original structure. We use three examples (a prismatic composite beam, an isotropic homogeneous tapered beam, and a composite tapered beam) to demonstrate the capability of HBE and show its advantages over the MSG cross-sectional analysis approach. HBE can capture macroscopic behavior and detailed stresses due to non-prismatic geometry.The Heterogeneous Solid Element (HSE) is developed based on MSG to model a heterogeneous body as an equivalent solid element using an effective element stiffness matrix. HSE modeling includes homogenization, macroscopic global analysis, and dehomogenization to recover local strains/stresses. HSE avoids the local periodicity assumption for traditional multiscale modeling techniques for composite structures that compute effective material properties instead. Abaqus composite solid element and MSG-based traditional multiscale modeling are used to validate the accuracy of HSE. All example results show that HSE is more accurate in predicting global structural behavior and local strains/stresses.HBE and HSE provide a new concept for modeling aperiodic composite structures by modeling structures into equivalent beam or solid elements instead of beam properties of the reference line in 1D beam analysis or material properties of material points in solid structural analysis
Ambient Electrostatics of Ions and Charged Microdroplets Produced Via Nanoelectrospray Ionization
No full textMass spectrometry, the science and technology of ions, owes much of its current popularity to the development of electrospray ionization. The development of electrospray ionization, along with its low flow-rate analog nanoelectrospray ionization, has increased the chemical space that can be investigated using mass spectrometers by orders of magnitude. While the interfacial chemistry of charged microdroplets that are generated by nanoelectrospray has been studied in detail, the physics of their motion, particularly in the presence of an applied field at ambient pressures, remains relatively unexplored. In this dissertation, an increase in ion currents detected by a commercial triple quadrupole mass spectrometer is used to demonstrate that: (i) the orthogonal injection of counterions into an electrode assembly can compensate for space charge effects and enhance the sampling of charged microdroplets from a nanoelectrospray focused electrostatically under ambient conditions into the mass spectrometer; and (ii) the ease of ion evaporation from charged microdroplets may be elucidated for small molecules based on their relative transmission through an electrode assembly for the simultaneous ambient electrostatic focusing of two nanoelectrosprays. In each case, the development is characterized by using ion trajectory calculations in conjunction with experiments, using homebuilt devices designed and fabricated in-house as rapid prototypes via 3D printing. In the open air, charged microdroplets have low kinetic energies with a narrow energy spread. Despite these limitations, this dissertation demonstrates, through the electrostatic manipulation of charged microdroplets produced via nanoelectrospray ionization, that a better understanding of the physics of moving charges in the open air can be used to increase the sensitivity of atmospheric pressure ionization
Learning and Decision Making Under Uncertainty
No full textIn practice, we often make decisions under uncertainties with known distributions or even without knowing distributions. This study explores decision learning and decision making in not-for-profit operations and supply chain management. We first study dynamic staffing under volunteer supply uncertainty, then explore how to dynamically balance uncertain supply with uncertain demand under lost sales, and finally study decision learning with limited data when distributions are unknown. We provide a brief description of the results obtained from the specific problems considered in this study.The dynamic staffing problem under volunteer supply uncertainty is explored in Chapter 2. We model a finite-horizon staffing problem in nonprofit organizations making hiring and assignment decisions for paid workers and volunteer, given a budget constrain, a capacity constraint, and uncertainties of volunteer supply and part-time worker turnover. Although the optimal staffing policy is computationally challenging to identify in general, we show that an intuitive prioritization assignment policy for all staff and a simple hire-up-to policy for part-time workers can be conveniently applied and close to optimal. Based on the theoretical properties of the optimal policy, we further suggest two easy-to-implement heuristics, both of which have low relative optimality gaps. We also provide performance lower bounds of both heuristics.The dynamic problem of balancing uncertain supply with uncertain demand under lost sales is explored in Chapter 3. We study the dynamic inventory replenishment and product pricing policy aiming to mitigate both supply uncertainty and demand loss. Since the dynamic planning problem is highly non-concave and thus intractable, we propose an approach that focuses on a class of intuitively appealing and practically plausible policies that require the amount of stock allocated for meeting the demand to be increasing and the product price to be decreasing in the available inventory level. We show that, under general conditions for the stochastic supply and demand functions, over a restricted monotone policy class, the dynamic problem become a concave optimization problem . We further reduce the restricted class to a refined class which can be easily computed, and appropriately selected refined policies produce optimal or close-to-optimal profits.The decision learning with limited data problem is explored in Chapter 4. We study how to utilize the data from related systems for decision making with limited data, which underscores the role of domain knowledge, the statistical similarity among the related systems and the structural relationships between inputs and outputs. When a related system has ample data, we demonstrate, through the application of newsvendor systems, that transfer learning can improve decision performance in the focal system, and cross-learning solutions can significantly improves the performance of the focal system over the transfer-learned solution and are asymptotically optimal. When there are multiple related systems with limited data, we transform the data from different systems to create a generic stochastic environment for the decision making problem, and show that the derived co-learning solution is asymptotically optimal for each involved system, as well as the aggregate system
Enhancing Cybersecurity of Unmanned Aircraft Systems in Urban Environments
No full textThe use of lower airspace for air taxi and cargo applications opens up exciting prospects for futuristic Unmanned Aircraft Systems (UAS). However, ensuring the safety and security of these UAS within densely populated urban areas presents significant challenges. Most modern aircraft systems, whether unmanned or otherwise, rely on the Global Navigation Satellite System (GNSS) as a primary sensor for navigation. From satellite navigations point of view, the dense urban environment compromises positioning accuracy due to signal interference, multipath effects, etc. Furthermore, civilian GNSS receivers are susceptible to spoofing attacks since they lack encryption capabilities. Therefore, in this thesis, we focus on examining the safety and cybersecurity assurance of UAS in dense urban environments, from both theoretical and experimental perspectives.To facilitate the verification and validation of the UAS, the first part of the thesis focuses on the development of a realistic GNSS sensor emulation using a Gazebo plugin. This plugin is designed to replicate the complex behavior of the GNSS sensor in urban settings, such as multipath reflections, signal blockages, etc. By leveraging the 3D models of the urban environments and the ray-tracing algorithm, the plugin predicts the spatial and temporal patterns of GNSS signals in densely populated urban environments. The efficacy of the plugin is demonstrated for various scenarios including routing, path planning, and UAS cybersecurity.Subsequently, a robust state estimation algorithm for dynamical systems whose states can be represented by Lie Groups (e.g., rigid body motion) is presented. Lie groups provide powerful tools to analyze the complex behavior of non-linear dynamical systems by leveraging their geometrical properties. The algorithm is designed for time-varying uncertainties in both the state dynamics and the measurements using the log-linear property of the Lie groups. When unknown disturbances are present (such as GNSS spoofing, and multipath effects), the log-linearization of the non-linear estimation error dynamics results in a non-linear evolution of the linear error dynamics. The sufficient conditions under which this non-linear evolution of estimation error is bounded are derived, and Lyapunov stability theory is employed to design a robust filter in the presence of an unknown-but-bounded disturbance
Weighted Curvatures in Finsler Geometry
No full textThe curvatures in Finsler geometry can be defined in similar ways as in Riemannian geometry. However, since there are fewer restrictions on the metrics, many geometric quantities arise in Finsler geometry which vanish in the Riemannian case. These quantities are generally known as non-Riemannian quantities and interact with the curvatures in controlling the global geometrical and topological properties of Finsler manifolds. In the present work, we study general weighted Ricci curvatures which combine the Ricci curvature and the S-curvature, and define a weighted flag curvature which combines the flag curvature and the T-curvature. We characterize Randers metrics of almost isotropic weighted Ricci curvatures and show the general weighted Ricci curvatures can be divided into three types. On the other hand, we show that a proper open forward complete Finsler manifold with positive weighted flag curvature is necessarily diffeomorphic to the Euclidean space, generalizing the Gromoll-Meyer theorem in Riemannian geometry
A Mixed-Methods Investigation of the Implications of Substance Use Disorder Stigma for Justice-Involved Youth
No full textObjectives: Compared to youth without justice-involvement, justice-involved youth are more likely to experience substance use disorders. Yet, few justice-involved youth receive appropriate, evidence-based treatment for substance use disorders. Although there are numerous barriers to the accessibility of appropriate treatment, research also suggests that it is difficult to engage justice-involved youth in treatment even when it is available and accessible. It is possible that substance use disorder stigma, or negative attitudes towards youth with substance use disorders, may contribute to low treatment accessibility, and make it more difficult for justice-involved youth to engage with available treatment. Few researchers have examined substance use disorder stigma among this population. The purpose of this study was to 1) explore the nature of substance use disorder stigma among justice-involved youth, at multiple ecological levels and 2) examine the role of substance use disorder stigma in limiting the accessibility of and engagement in treatment and justice-involved youth’s engagement in treatment.Methods: Participants (n = 44 youth-guardian dyads) were referred to the study by juvenile probation departments in two Indiana counties. In addition, 66 system personnel participants who work with justice-involved youth with substance use disorders were recruited from community mental health centers and juvenile probation departments in rural and suburban Indiana counties. All participants completed survey measures of substance use disorder stigma and familiarity with substance use; youth-guardian dyads also provided information about the youth’s substance use history and treatment utilization. A subset of participants (n = 9 youth, 11 guardians, 12 system personnel) completed qualitative interviews, providing perspectives on substance use disorder stigma and the role of stigma in discouraging treatment. Using analysis of covariance, multiple regression analyses, and qualitative grounded theory analysis, I explored the nature of stigma toward justice-involved youth with substance use disorders and examined the impact of stigma on treatment accessibility and engagement.Results: For aim 1, as hypothesized, public stigma (assessed by survey data) varied significantly according to participant role and specific substance, with guardians endorsing greater stigma than system personnel. All participants expressed greater negative emotions towards youth with opioid use disorder compared to marijuana use disorder. Interview data revealed particularly nuanced attitudes about marijuana use. Contrary to expectations, youth reported little self-stigma. Both youth and guardians described limited knowledge of problematic substance use. For aim 2, interview data suggests that youth and guardians may identify more stigma associated with seeking treatment for problematic substance use than with using substances. All participants reported that perceived stigma has improved in recent years, and that youth feel more comfortable discussing their substance use. However, guardians identified family attitudes about behavioral health treatment as negatively impacting engagement among youth. In addition, system personnel reported that stigma continues to limit the accessibility of youth SUD treatment.Discussion: Youth endorsed lower than expected levels of self-stigma with no difference by primary substance type; this may have been affected by youth’s limited understanding of problematic substance use and lower than expected heterogeneity in substance use type among participants. Consistent with prior research, self-stigma was directly related to the severity of mental health symptoms. The high prevalence of public stigma among guardians of JIY with SUDs suggests that parents and guardians would benefit from interventions to better support their caregiving experiences