Treasures @ UT Dallas

Welcome to Treasures @ UT Dallas Institutional Repository, established in 2010. Treasures is a resource for our community to showcase, organize, share, and preserve research and scholarship in an Open Access repository.


Recent Submissions

Transfer Learning and Uncertainty Quantification in Natural Language Processing for Political Science and Cyber Security
(2023-08) Hu, Yibo; Khan, Latifur; Makris, Yiorgos; Ouyang, Jessica; Brandt, Patrick T.; Du, Xinya
Recent advancements in Natural Language Processing (NLP) driven by pretrained language models have revolutionized various fields reliant on large-scale text-based research through transfer learning. This dissertation presents efficient, reliable computational NLP applications to address real-world challenges, with a focus on political science, cyber security, and uncertainty quantification. The dissertation begins with interdisciplinary research in political science, where advanced NLP models are developed to track and analyze dynamics related to global political conflict. The creation of ConfliBERT, the first domain-specific sociological language model, enables improved performance on 18 downstream tasks, particularly in scenarios with limited data availability. Moreover, by leveraging transfer learning and existing expert knowledge, specific tasks such as political event extraction and classification are further optimized. One approach called Confli-T5 is a text generation model that augments labeled data by in- corporating achievable templates derived from political science knowledge bases. Another technique introduced is the Zero-Shot fine-grained relation classification model for PLOVER ontology (ZSP), which eliminates the need for labeled data by relying solely on an annotation codebook to classify intricate interactions between political actors. These strategies combine the power of transfer learning with domain-specific expertise to reduce the dependence on extensive labeled data, making them valuable tools in the field. In the field of cyber security, text generation techniques are employed for cyber deception, generating multiple fake versions of critical documents to deter malicious intrusion. A context-aware model called Fake Document Infilling (FDI) addresses the limitations of existing approaches by considering contextual awareness. FDI produces highly believable fake documents, protecting critical information and deceiving adversaries effectively. Finally, uncertainty quantification techniques are explored to enhance the reliability of NLP models in such interdisciplinary or cross-domain applications. A novel model, BERT-ENN, employees evidential theory to quantify multidimensional uncertainty in the data and calibrate uncertainty estimation in text classifiers. This approach achieves state-of-the-art out-of-distribution detection performance, thereby improving the reliability of NLP models.
Risk-based Motion Planning and Control for Robotic Systems
(2023-12) Safaoui, Sleiman; Summers, Tyler; Kang, Gu Eon; Spong, Mark W.; Koeln, Justin; Ruths, Justin; Vinod, Abraham P.
A robot autonomy stack usually consists of several modules that enable it to perceive the environment and decide how to interact with it to achieve a desired task. At the heart of this stack are the motion planning and control modules. The motion planning module is generally responsible for decision making and generating a plan for the robot to follow, such as determining how an autonomous car should drive around pedestrians and other vehicles. The control module computes a finer sequence of control actions that can be issued to the actuators to operate the robot. One issue that plagues robot motion planning and control is the effect of uncertainty, of which there are different types, on the system. This includes unknown and unmodeled disturbances that affect the system such as noise, aerodynamics, or simplified dynamics models. However, addressing these uncertainties is non-trivial and often requires a trade-off between accounting for the uncertainty accurately and the tractability of solving the problems. This dissertation develops risk-based solutions for a few robot motion planning and control problems. The contributions of the dissertation are categorized into four main types. The first part addresses control design with complex spatio-temporal requirements under uncertainty. An optimization-based control algorithm is designed to guarantee the completion of the requirements when the robot dynamics are affected by process noise. The second part addresses sampling-based motion planning under uncertainty. RRT*, a famous motion planning algorithm in robotics, is considered and risk-aware variants of it are developed to account for process and measurement noise affecting the robotic system. The third part addresses a limitation of learning-based planning approaches with an application to multi-agent motion planning. A reinforcement learning (RL) framework is considered for learning policies then an optimization-based module, called a safety filter, is proposed to enforce collision avoidance as hard constraints, which learning algorithms cannot do. The safety filter is designed to handle process, state, and measurement noise. Finally, the fourth part addresses data-driven planning in dynamic and uncertain environments. This assumes that the robot has access to some future predictions of the obstacles in the environment, such as where they may be in the next few seconds. A safety filter is then developed using these sample predictions to plan a safe trajectory for the robot. In several sections, uncertainties whose distribution is unknown, which is generally the case, are considered and addressed using the concept of distributionally robust optimization (DRO) to develop solutions that guarantee safety or the successful completion of the task despite the lack of knowledge of the underlying distribution. Throughout, examples are provided to emphasize and clarify core concepts, and simulations and physical experiments are performed to demonstrate the efficacy of the developed solutions.
Student Haiku Competition 2024
(Eugene McDermott Library, 2024) Aries De Joy Uy, Lester; Raghavaraiu, Nikhitha; Ramsey, Carson
Low Noise Integrated Circuits and Systems Using Nano-Scale MOSFETs and Intelligent Post-Fabrication Selection
(December 2021) Yelleswarapu, Venkata Pavan Kumar; O, Kenneth K.; Venkatesan, Subbarayan; Henderson, Rashaunda; Ma, Donsheng Brian; Makris, Yiorgos
Recent advances in integrated radio design have enabled many applications such as wearable healthcare, 5G communication, and beyond 5G or 6G applications for ultra-high data rate communications, high-resolution imaging, sensing, and spectroscopy. All these applications require low noise radio transceivers for achieving high performance. For example, applications requiring high data rate and higher order modulation schemes need to achieve high signal to noise ratio (SNR) and therefore a low noise figure to maintain a low bit-error rate (BER). In addition, noise phenomena like jitter and phase noise can impact the critical parameters like maximum achievable data rate and energy efficiency. This research aims to improve the noise performance of integrated circuits and systems through intelligent post-fabrication selection of an array of nanoscale transistors sized near the minimum in CMOS processes. A phase noise reduction technique in LC Voltage Controlled Oscillators (VCO’s) is demonstrated by post-fabrication selection of a subset of an array of near minimum-size cross-coupled transistor pairs with reduced low frequency noise and thermal noise. The technique reduces the phase noise by taking advantage of the fact that when transistor dimensions are reduced, the low frequency noise and thermal noise vary significantly. Applying an intelligent post-fabrication selection process using a genetic algorithm, the lowest phase noise of -122 dBc/Hz, -127 dBc/Hz, -137.5 dBc/Hz at 600-kHz, 1-MHz, and 3-MHz offsets, respectively from a 3.8-GHz carrier has been measured. The VCO prototype was fabricated in a 65-nm CMOS process and dissipates 7 mW of DC power. The maximum figure of merit (FoM) including phase noise, carrier frequency and power consumption is 191 dBc/Hz and the figure of merit including the VCO core area, FoMA is 207 dBc/Hz. A technique is demonstrated to reduce both the in-band and out-of-band phase noise of a 4-GHz Integer-N PLL by employing an array of individually selectable cross-coupled pairs formed using near minimum-size transistors in an LC VCO and intelligent post-fabrication selection. By reducing both the in-band and out-of-band phase noise, the overall integrated phase jitter in a frequency synthesizer can be minimized. Applying an intelligent post-fabrication selection process, the lowest phase noise of -72 dBc/Hz at 30-kHz offset, -106 dBc/Hz at 300-kHz offset, -121.8 dBc/Hz at 1-MHz offset, and -132.5 dBc/Hz at 3-MHz offset, respectively from a 4.01-GHz locked carrier has been measured. The integrated rms jitter from 100-kHz to 100-MHz offsets is 440 fs. A mixer-first downconverter employing an array of passive mixers formed using near minimumsize transistors and intelligent post-fabrication selection achieves a double sideband noise figure of 4.2 dB at RF of 6 GHz, which is the lowest at 6 GHz for CMOS mixer-first downconverters. The downconverter is fabricated in 65-nm CMOS and demonstrates out-of-band IIP3 and IIP2 of 25 dBm and 65 dBm, respectively at 80-MHz IF, while dissipating 11.5 mW. Post-fabrication selection is performed by a genetic algorithm which takes ~17 generations to converge to the combinations exhibiting the lowest noise.
Causarum Cognitio: the Architecture, Collections, and Social Agency of Three American Athenaea: Redwood, Boston, and Caltech
(December 2021) Curry, Virginia; Schulte, Rainer; Gooch, John; Channell, David; Schlereth, Eric; Schich, Maximilian
Is the athenaeum an adaptable concept in the twenty-first century university environment? What evidence exists to conclude that it contributes to a discursive community? This dissertation explores the legacy of the concept of the athenaeum in America and examines the organically formed social circles who share an interest in continuing discourse, often within multiple disciplines, and who contribute to their communities by modeling habits and behaviors reflecting their desire for improvement of themselves and their communities. From before and since our nation’s founding, the societies of the American Athenaeum have served as community-organized intellectual and artistic hubs, providing access to information, pursuing thought-provoking discourse, and applying their aggregate knowledge resources as agency for social change while presenting the most inspirational architecture, lectures, artistic performances, and collections to their communities. I focus on the eighteenth century Redwood Library and Athenaeum of Newport, Rhode Island, the nineteenth century Boston Athenaeum, and the twentieth century Caltech Athenaeum. The newest of these, Caltech Athenaeum, has been in service over one hundred years, and the oldest, the Redwood Library and Athenaeum, has been in service to its community continuously over 300 years.