116964 research outputs found
Sort by
Investigating the effects of a machine-learning powered framework on interactive engineering design
Get PDFIn a collaborative effort involving re:3D, Austin Habitat for Humanity (AHFH), The University of Texas at Austin, and other partners, our team is developing a system that provides an opportunity for non-professionals to create their own custom-made, 3D-printed furniture and home goods by upcycling plastic waste sourced from the local community. Within this process, our team is also exploring the integration of an AI-powered interactive design framework that allows customers to lead the design process of their own structurally sound home goods without any prior engineering or design expertise. The recent development of machine learning algorithms has led to the implementation of human-AI (artificial intelligence) co-creative systems in various sectors with the potential to transform the interactive design process. However, the way humans interpret and interact with machine learning tools and the effect of the tools on human-guided design processes are not fully understood yet. Recent research has attempted to quantify characteristics of generative AI models by conducting various user studies, but the focus has been on characterizing the model rather than analyzing the interaction patterns and behavior of users. Also, the user studies often employed simple GUIs that perform one task only, which is not sufficient to model realistic design processes. In this thesis, a preliminary user study is conducted with a fully co-creative environment in which all the searching and editing is driven by the user with the assistance of an automated system built on AI and specifically, design representations generated by conditional generative adversarial networks (cGANs). As users design their own furniture and recreate a specific piece of furniture using the framework, we attempt to emulate an organic, iterative, and user-driven design process. Furthermore, we investigate user strategies and behavior when solving different design problems, changes in user preferences and tradeoffs associated with interaction with the design framework, and the effect of AI-powered searching and editing functions on users’ convergence to a final design.Mechanical Engineerin
Improving searches for gravitational waves and searching for exotic objects
Get PDFThis dissertation explores advancements made in gravitational wave searches, with a focus on techniques and methods that improve detection sensitivity as well as the search for novel sources of gravitational waves in the form of exotic objects, such as primordial black holes and boson stars. It is divided into 7 chapters, where chapter 1 introduces foundational concepts, including the derivation of gravitational waves and their detection through interferometric observatories, such as LIGO, VIRGO, and KAGRA (LVK). The subsequent subsections in chapter 1 provides brief introductions to specialized topics such as the gravitational wave detection pipeline GstLAL, the statistical inference framework iDQ, and the search for exotic objects. These introductions each lead into discussions of my contributions to two related papers that are found in accompanying chapters. Chapters 2 and 3 are GstLAL based projects which discuss improvements to GstLAL's search sensitivity through an improved ranking statistic and the implementation of a neural network that can lead to faster parameter estimation. Chapter 4 and 5 are iDQ papers which discuss the overall improvements to the framework made between the third and fourth observing runs of LVK and a study that demonstrates how the inclusion of data quality in the GstLAL pipeline can lead to increased search sensitivity. Finally, Chapters 6 and 7 are exotic object search papers which discuss the implementation of a waveform that can be used to search for exotic objects and a novel method that allows for the rapid generation of gravitational wave template banks.Physic
Master's thesis recital (violin)
No full textViolin sonata no. 6, op. 27 / Eugène Ysaÿe -- Sonata in A major for violin and piano / César Franck -- Selections from Porgy and Bess / George Gershwin ; transcriptions by Jascha Heifetz.MusicName of supervisor not provide
Reactor core from phosphate ore
Get PDFThis thesis addresses the processes necessary to extract uranium from phosphate ore and convert it to reactor fuel which can be used in a 1,000 MWe PWR. This includes uranium extraction from phosphate ore, conversion, enrichment, deconversion, and fuel fabrication. Uranium demand has increased over the past four years which will likely influence whether phosphoric acid production companies restart uranium extraction activities. Environmental and political drivers will also almost certainly have a significant influence on this market, in addition to economic drivers. This thesis provides a summary of historic solvent extraction systems (D2EHPA-TOPO, OPPA, OPAP) and mentions a few newer systems with higher distribution ratios (e.g. BDEHCNP) that will likely be candidates for future processes. Ion exchange also shows significant promise as a possible replacement technology. An evaluation of capital and operating costs indicates that in 2024 solvent extraction would have a total cost of [dollar sign]59-[dollar sign]83/lb U3O8 and ion exchange would cost [dollar sign]46-[dollar sign]76/ lb U3O8. Environmental and health effects are also addressed along with national security and proliferation concerns.Mechanical Engineerin
Large language models are capable of offering cognitive reappraisal, if guided
Get PDFLarge language models (LLMs) have offered new opportunities for emotional support, and recent work has shown that they can produce empathic responses to people in distress. However, long-term mental well-being requires emotional self-regulation, where a one-time empathic response falls short. This work takes a first step by engaging with cognitive reappraisals, a strategy from psychology practitioners that uses language to targetedly change negative appraisals that an individual makes of the situation; such appraisals is known to sit at the root of human emotional experience. We hypothesize that psychologically grounded principles could enable such advanced psychology capabilities in LLMs, and design RESORT which consists of a series of reappraisal constitutions across multiple dimensions that can be used as LLM instructions. We conduct a first-of-its-kind expert evaluation (by clinical psychologists with M.S. or Ph.D. degrees) of an LLM's zero-shot ability to generate cognitive reappraisal responses to medium-length social media messages asking for support. This fine-grained evaluation showed that even LLMs at the 7B scale guided by RESORT are capable of generating empathic responses that can help users reappraise their situations.Linguistic
Manufacturing immersion gratings, evolving circumstellar disks, and the archive in-between
Get PDFThe grating group at the University of Texas at Austin has spent decades developing and refining the process of manufacturing silicon immersion gratings for use in high-resolution near-infrared (NIR) spectrographs. The success of our manufactured gratings relies heavily on contact lithography, exposing a patterned photo-mask in direct contact with a photo-resist coated substrate using a highly uniform UV light source we call the UV Exposure System. In Chapter 1, I discuss the design and testing of an improved high uniformity UV Exposure system, which reaches 6.25% non-uniformity over a 130 mm diameter while preserving a high-intensity power throughput. The Immersion GRating INfrared Spectrometer (IGRINS) is a high resolution (R ~ 45,000) NIR spectrograph that simultaneously covers the H- & K- bands (1.45-2.45µm) in a single pointing and employs one of our high-quality silicon immersion gratings. After being commissioned at McDonald Observatory in 2014, IGRINS has had continued success at the Lowell Discovery Channel Telescope (from 2016-2018) and Gemini South (2018-2024), producing over 25,000 H- & K-band spectra from 10 years on sky. In Chapter 2, I introduce The Raw and Reduced IGRINS Spectral Archive (RRISA), which provides streamlined public access to IGRINS raw and reduced data products featuring over 3,500 unique observations of both science targets and A0V standards. RRISA not only includes access to all of the raw and reduced IGRINS data products but also cross-matches all of the IGRINS targets with popular catalogs such as SIMBAD, 2MASS, APOGEE, PASTEL, Gaia DR2, and Gaia EDR3. In Chapter 3, I use RRISA to build a sample of 108 Taurus young stellar objects (YSOs) to study molecular hydrogen (H₂) emission from the disks of YSOs. I present our method for estimating the radial velocities for each of our YSOs using SAPHIRES and compare the results to existing radial velocity measurements featured in SIMBAD and APOGEE. Finally, I discuss the future characterization of the YSOs in our sources to separate the various H₂ excitation mechanisms and our progress towards building a simple H₂ emissivity model.Astronom
The check
Get PDFThe Check is a full-length play about an alternative America where Black Americans are finally receiving reparations after four hundred years. At the center of the play is a Black man, named C.J., who gets caught between using the reparation check to help his family or on a business investment promising high returns. It is a play about reparations on the macro and micro levels: what a country owes to the people who helped build it but are continuously being treated as second-class citizens, what a husband owes his wife, what a parent owes their child, what a child owes their parent, what a friend owes a friend, and what the living owe the dead.Writin
Peeling the layers of visibility regimes : adopting a critical-communicative lens to examine onion supply chain in India
Get PDFVisibility has become a powerful force for fostering accountability and transparency, amplifying voices, and uncovering hidden structures. It determines what is seen or obscured, recognized or neglected, and acknowledged or suppressed within a given system. In particular, enhancing visibility in supply chains (from farm to fork) has become a central concern and primary goal for academics and practitioners alike. In fact, all-encompassing visibility is deemed a panacea for mitigating a variety of supply chain challenges. Given the ever-increasing significance, the overarching goal of this study is to examine how visibility is communicatively produced in an agricultural supply chain.
In supply chain management scholarship, visibility is typically operationalized from an instrumental lens, defining it as the ability to access real-time information about the location, status, and condition of goods and processes across the supply chain, enabling transparency, efficiency, and proactive decision-making. Stepping beyond this conventional and instrumental conceptualization of visibility, this study conceives of visibility as a layered, communicative, sensory, relational, and power-laden phenomenon. Focusing attention on onion supply chains in India, this study seeks to critically conceptualize visibility regimes, referring to the ways in which visibility shapes practices, processes, and activities embedded in the daily lives of supply chain actors and is in turn shaped by them. To do so, this project investigates supply chain visibility by proposing two research questions—how do the experiences of the onion supply chain actors help construct the visibility of the chain? And how is visibility constituted in the onion supply chain? These two questions are aimed at understanding visibility at different levels in the onion supply chain and to paint a more comprehensive picture that does not reduce visibility to datafied or informational forms. To address these questions, an ethnographic study was conducted in the onion-belt region of India.
First, this study demonstrates the value of incorporating the voices, experiences, and perspectives of actors in the practice of supply chain mapping to encapsulate the underlying power dynamics between actors. The mapping process revealed that the onion supply chain actors have spheres of control in which dialectical tensions emerge and shape the agency of actors in different ways. The boundaries shape their range of visibility of the chain which in turn impacts their participation and power in the chain.
The second research question peels the layers of visibility regimes to its core to examine how visual dimensions are augmented by other senses, such as olfactory, tactile, and auditory to offer a sensory conceptualization of visibility. These combinations of senses operate on a day-to-day basis when actors manually handle the commodity and engage in visibilization processes such as presentation, monitoring/inspection, orchestration, categorization, aggregation, and datafication. These processes manifest in each of the five key steps associated with onion auctions, with analog and digital forms of datafication present in four out of the five steps. These visibilization processes are sociomaterial and demonstrate the importance of centering the commodity when examining visibility in an agricultural supply chain. This study critiques the notion that visibility regimes in supply chains hinge upon advanced digital platforms. In traditional chains such as the onion chain, actors primarily rely on sensory engagement with the commodity (size, shape, smell, color, texture, layers, layers, firmness of onions), and because existing technology is yet to capture these nuanced and complex sensory interactions, visibility regimes emerge and are perpetuated through mundane social activities involving the commodity. Thus, visibilization occurs in and of the chain and cumulatively feeds into the visibility regimes that foreground certain actors while relegating others in the chain.Communication Studie