26400 research outputs found
Sort by
Neural Networks for Decisions Under Uncertainty
Get PDFNeural networks are used in many real-world applications, ranging from classification tasks to medical diagnostics. For each task, a neural network is typically able to make predictions due to its ability to extract meaningful patterns from processing large amounts of data. Thus, given the increases in available data in recent decades, the performance of neural networks in making accurate predictions has greatly increased. However, this data often comes with ingrained uncertainties due to measurement errors or the inherent variability of individual data points. Neural networks can learn despite the errors in the overall data, but what if we want them to consider the uncertainty present in individual data points?
This master\u27s thesis explores whether neural networks can make reliable predictions given inputs with embedded uncertainties. We further expand this idea into two main areas of exploration: the performance of neural networks with uncertain data represented with real values and the design of neural networks that can model and process uncertain data directly throughout their framework. We test the first idea in robot localization scenarios, which produce uncertain data due to the errors present in measurements but nevertheless require reliable results for many applications, such as autonomous vehicles. For the second area, we explore methods for propagating uncertainty throughout the neural network learning process. Such neural networks represent and maintain uncertainties through their computations, potentially creating more accurate and robust predictions under uncertainty. Methods developed in both exploration areas are tested and compared against existing techniques, such as interval computations and traditional dense neural networks, providing mixed results in their overall reliability and accuracy
The Lack of Representation of Mexican Women in Films
Get PDFRepresentation in film lacks the meaningful exposure of Mexican women, the lack of roles available tend to rely on the stereotypes of Mexicans therefore excluding room for Mexican women to be represented authentically. In this paper I discuss the lack of representation that Mexican women have in films, identifying that in roles of superheroes, especially women of color do not compare to the many white superhero films. To discuss representation of Mexicans in film, I used The Perfect Game (2009) and McFarland, USA (2015), films that share meaningful representation of Mexicans. While also using examples of representation in superhero films like Black Panther (2018) and Shang Chi and the Legend of the 10 Rings (2021), both films share the representation of two minorities that have not been seen as leads in a superhero film. Finally, I discuss how the attempts of meaningful representation of Blue Beetle means to the representation of more Mexicans in films, even women, and how more could have been done in order to be successful with showcasing representation
Deciphering Rodent Behavior: Unraveling Complex Decision-Making Through High-Throughput Data Analysis In Behavioral Neuroscience
No full textUnderstanding the effects of substance use on decision-making is critical for both clinical andtranslational research. In this thesis, I present two complementary studies that explore decision- making behavior in rodents with a focus on alcohol consumption and novel experimental frame- works. The first study examines the impact of acute alcohol consumption on decision-making in a behavioral approach-avoidance task. Rats were presented with a concurrent choice of consuming alcohol and sucrose mixtures across four distinct concentrations, allowing for the analysis of ap- proach or avoidance behavior. Sex-dependent differences emerged, with male rats demonstrating a significantly higher tendency to approach higher alcohol concentrations, while female rats showed minimal changes in their choice behavior. This novel protocol highlights the individual variability in decision-making under alcohol exposure and provides a method for identifying abnormalities that could predict vulnerability to alcohol use disorders. The second study introduces the REward-COst in Rodent Decision-making (RECORD) sys- tem, an automated, high-throughput framework designed to study decision-making across multiple reward and cost levels. By integrating 3D-printed arenas, custom electronic hardware, and soft- ware, RECORD enables the analysis of complex decision-making behaviors without the need for food or water restriction. Using test cases involving oxycodone self-administration and alcohol consumption, we reveal how individual and group behavioral heterogeneity can be quantified, of- fering new insights into perturbations in decision-making. Together, these studies provide innovative tools and insights for studying decision-making un- der substance use, with the potential to inform research on psychiatric disorders and addiction vulnerability
Production Of Labiodental Fricative /v/ In Heritage Speakers Of Spanish
No full textAbstractThe present study investigates the factors that influence Heritage Spanish speakersâ?? production of the allophone [v]â??namely its corresponding orthography ( vs. ), the phonological context in which it is produced (VCV vs. NCV, where V = vowel, C = consonant, N = /n/), and participantsâ?? language dominance (Spanish vs. English). More specifically, this study seeks to determine whether, for these Spanish speakers, [v] contrasts with /b/ (hence, /v/) or is an allophone of /b/. Standard Spanish does not have /v/, but it has â??an orthographic form of /b/, which in turn has (at least) two allophonic realizations: [b] after nasals and pauses, and [β] elsewhere. It has been reported that [v] occurs in some Spanish dialects in contact with languages that have /v/ (e.g., English; Hualde & Colina, 2014), especially in Spanish-English cognates (Davidson, 2019). However, the phonological status of [v] in these Spanish speakersâ?? productions is unclear. Forty native Spanish speakers living in the US who varied in their dominance in Englishâ??as determined by the Bilingual Language Profile questionnaire (Gertken, Amengual, & Birdson, 2014)â??completed a picture-naming task, a spelling task (to ensure that they knew how to spell the Spanish words tested), and a read-aloud task. The target words started with an orthographic or and were preceded by a feminine determiner (VCV, where [β] is expected) or masculine determiner (NCV, where [b] is expected) (e.g., un bote, una bota, un velo, una vela). All words were disyllabic and stressed on the penultimate syllable. Participants were audio- and videorecorded while producing speech, and the words were coded for whether a [v] was produced based on the visual articulatory information provided by the videorecordings (Vergara, 2010). The results revealed a marginal interaction between task and orthography, with a stronger effect of orthography on [v] productions in the read-aloud task than in the picture-naming task. The results also revealed a significant three-way interactions (with task as a fixed effect) between orthography, phonological context, and language dominance, with the effect of orthography being stronger in VCV contexts than in NCV contexts, and with that interaction being stronger in English-dominant speakers than in Spanish-dominant speakers. The numerical results showed that the great majority of [v] productions were found in the presence of orthographic and in VCV contexts, where [β] is expected. This was true for both Spanish-dominant and English-dominant speakers, but the effect was more accentuated in English-dominant speakers. These results strongly suggest that Heritage Spanish speakers treat [v] as an allophone of /b/ (similarly to [β]) in the presence of an orthographic in the target words. We propose that this orthographic effect is perceptually driven, revealing an important interaction between orthographic and phonological representations of bilinguals
The Influence Of Pitch Type (fastball, Curveball, And Change-Up) On The Contralateral Trunk Tilt In Adolescent Pitchers
No full textBaseball participation has grown, with over 15.59 million players aged six and older in 2021. Pitching relies on effective coordination of the kinetic chain to transfer forces and generate ball velocity. While trunk rotation during pitching has been studied, contralateral trunk tilt across pitch types remains underexplored. This study investigated differences in hand velocity, stride length, and contralateral trunk tilt between fastball, curveball, and change-up pitches in 16 adolescent pitchers ages 15-18. Motion data were collected using a seven-camera system interfaced with Theia 3D. Repeated measures ANOVA revealed a significant difference in hand velocity between fastball and change-up (p = 0.035), with no significant differences in stride length (p = 0.236) or trunk tilt (p = 0.782). These findings indicate adolescent pitchers maintain consistent mechanics across pitch types, with hand velocity variations reflecting efforts to disguise pitch type. Efficient energy transfer through the kinetic chain is critical for optimizing performance, further research in live settings is needed to confirm these results
Monitoring Of Laser Soldering Process In Hybrid Additive Manufacturing
No full textAdditive manufacturing (AM) has been a maturing technology that has expanded over the course of the last decade in major advancements especially in accessibility, process monitoring, modeling, and stock material design. As its industry use is growing, the demand for added improvements such as multi-functionality increases, such as with a hybrid system that allows for modification of printed parts with different electrical components like the Foundry Multi-3D system. The system\u27s purpose is to create AM parts with multifunctional electrical components with the vision to expand to other areas of interest in manufacturing such as ISAM. This work explores the use of one process in the Foundry system, laser soldering, by utilizing spectral pyrometry in a feedback control system that can be used to control temperature and detect defects. Laser soldering in hybrid additive manufacturing has yet to be fully studied regarding process development and control, but in the case of printed circuit board industry this process has been studied. This work utilizes the Hybrid AM process to create specimens containing 22- and 28-gauge wire in top-down solder joints. These joints were created using a PID temperature control system to reach a temperature of 280°C followed by a cooldown stage. During the soldering process, both laser output and temperature data were collected along with camera video and its respective image processing data for burning. Through temperature vs power plots, signal to noise ratios, and frequency analysis, soldering events can propagate through disturbances in temperature. The temperature vs power relationships show that the power requirements for 22-gauge wire is nearly twice of that of the 28-guage (7.94 Watts and 4.07 Watts respectively), and due to the design where joint locations have different heat paths, there is over a 1-Watt standard deviation for both gauge wires. From power requirements and thermal capacitance, it was identified the process can be scalable. The effect of thermal capacitance is seen through heating and cooling phases where some hysteresis occurs, this is more pronounced from the 22-gauge wire as it has larger thermal mass. Signal to noise ratios shows high amplitude peaks can relate to burning and wire movement, the difference between can be identified through the temperature where burning is seen more than 300ºC. The frequency analysis showed subharmonics peaks to the control frequency (333 Hz), where 29 Hz and 58 Hz showed relatively high peak disturbances. A narrow pass filter was applied to the temperature signal to identify possible disturbance peaks between both frequencies, and possible trends that occurred. From the 29 Hz signal, there were some peaks identified for burning, wetting, and phase change, however the peaks amplitude was inconsistent through other joints. This work identified that further control implementation to account for wire size heating requirements and thermal capacitance (through lead-lag functions) would be necessary to scale to other wire sizes. The signal to noise ratio and frequency domain can be used to identify certain events, however further refinement or advanced techniques would be necessary for better confidence in event detection
Interest Groups and Central Banking
No full textThis chapter offers an alternative approach to the public interest theory of central banking.Instead of interpreting the history of central banking as attempts to solve market failures, we suggest that central bank behavior can be best explained by studying the constraints faced by members of various interest groups, including members of the central banks themselves. Instead of looking uniquely at how private decisions in commercial markets shape central banks’ decisions, we argue that central bank behavior is best predicted when applying the tools of economics to politics, i.e., when using Public Choice
Advancing Grid Modernization Through Data-Driven Resilience Modeling and Hosting Capacity of Distributed Energy Resources
No full textThis Ph.D. dissertation focuses on advancing the integration of distributed energy resources (DERs) through concepts surrounding their impacts on power system stability, resilience, and hosting capacity (HC). This dissertation addresses crucial topics in renewable energy deployment, transient fault response, and dynamic modeling. The work begins with the development of renewable energy source (RES) models tailored for offsetting residential heating, ventilation, and air conditioning~(HVAC) and commercial cooling loads. These models utilize solar photovoltaic (PV) and wind energy systems to produce scalable frameworks adapted across diverse climates and building types in application of a partial-load targeting methodology. The dissertation then transitions from the modeling of DERs to examining resiliency of electric power grids containing distributed PV deployment strategies when exposed to disturbances from singular faults to extreme weather events. This is done to highlight the stabilizing and recovery potential of networks with various penetrations of DERs. This concept is expanded by performing comparative analyses of centralized and distributed PV configurations to further detail deployment strategies against various resilience metrics to provide critical insight on their respective trade-offs in grid resilience and operational efficiency. The final phase of the dissertation transition to load-side assessments in hosting capacity~(HC) and its susceptibility to common short-circuit faults. This evolves into the development of a real-time HC framework leveraging OpenDSS and RT-Lab (OPAL-RT) in a bidirectional interface that allows user input to manipulate dynamic electric vehicle (EV) loads and stochastic fault injections into synthetic distribution networks and observe live HC evaluations based on voltage stability profiles of system feeders. The intellectual merit of this work lies in its contributions to grid modernization efforts through advancing the critical understanding of DER integration and its interplay with disturbance contingencies to ultimately seek methods and strategies for maintaining a resilient power grid. The major contributions of this dissertation are to investigate and uncover the interplay between DER integration and extenuating circumstances resultant of simple, moderate faults all the way to extreme weather disturbances. This involves (1) effective development of novel RES modeling concepts to target quantifiable load portions such as residential or commercial space conditioning elements; (2) transient stability analyses that demonstrate the role of distributed solar PV systems in mitigating fault impact as well as their ability to support system recovery; (3)~investigation into the primary trade-offs between centralized and distributed solar PV applications to provide insights into their respective impact on grid resilience; and (4) critical impact assessments on the effects of common short-circuit faults on distribution system HC with regard to voltage stability