Physical Sciences and Mathematics Commons^™

Learnable Structured Attention For Student-Aware Knowledge Distillation In Dense Prediction, Chen Liu

ETDs from 2020-2029

Dense prediction tasks, including object detection and semantic segmentation, require models to produce structured predictions and are widely used in real-world vision applications. Although deep networks have achieved strong performance on these tasks, their high computational cost limits deployment on resource-constrained systems such as autonomous-driving vehicles. Knowledge distillation (KD) addresses this issue by transferring knowledge from a large teacher model to a compact student model. However, existing distillation methods for dense prediction face greater challenges than those for classification due to the more complex task requirements. To overcome the challenges, this thesis presents a unified study of adaptive distillation for …

A Modular Saponin Adjuvant Platform: Development, Structure–Activity Relationships, And Applications, Di Bai

ETDs from 2020-2029

ABSTRACT Saponin-based vaccine adjuvants have attracted considerable attention owing to the clinical success of QS-21, a potent immunostimulant capable of inducing balanced humoral and cellular immune responses, yet its widespread application is limited by supply constraints, chemical instability, and dose-dependent toxicity. To address these challenges, we previously developed two semisynthetic saponin adjuvants, VSA-1 and VSA-2, derived from abundant Momordica saponins through a single-step modification, which exhibit QS-21–comparable immunopotentiating activity with improved accessibility and stability. Here, we first performed a systematic structure–activity relationship (SAR) study of the VSA series using MS I as the lead scaffold. Modulation of the side-chain length …

Investigation Of Air Emissions With A Local Composites Industry, Chris White

ETDs from 2020-2029

Workers at an office space raised complaints about a sweet-smelling odor directly outside their building. After an initial assessment of the surrounding businesses and their processes, it was determined that the workers were most likely exposed to styrene emissions from a nearby composites company. This investigation utilizes multiple weeks of direct readings with a photoionization device and active environmental sampling with a charcoal tube and pump to confirm the styrene exposure hypothesis. Weather trends were also recorded to determine correlation between weather and exposure to the sweet-smelling chemical. The direct readings found an exposure maximum of less than 2 ppm …

Statistical Methods For Mendelian Randomization Under Nonlinearity And Non-Normality, Mary Appah

ETDs from 2020-2029

Instrumental variable (IV) methods are widely used for estimating causal effects in observational studies where unmeasured confounding may bias traditional regression estimates. The core idea is to use a variable referred to as an instrument that is associated with the exposure of interest, independent of unmeasured confounders, and influences the outcome only through the exposure. While originally developed in econometrics, IV methods have been increasingly adopted in epidemiology and genetic research under the framework of Mendelian Randomization (MR), where genetic variants, most commonly single nucleotide polymorphisms (SNPs), serve as instruments. MR provides a powerful tool for investigating causal relationships between …

Amending Planting Media With Typha-Derived Biochar For Sustainable Horticulture, Elissa Touma

USF Tampa Graduate Theses and Dissertations

Constructed wetlands are widely used to improve water quality by removing excess nutrients from wastewater and stormwater. Nutrients are stored within wetland vegetation, particularly in fast-growing plants such as Typha. Harvesting vegetation provides an opportunity to permanently remove these nutrients while generating biomass suitable for beneficial reuse. Converting harvested biomass into value-added products such as compost and biochar can support nutrient recovery and sustainable biomass management. In addition, increasing concerns regarding the environmental impacts of peat extraction have created interest in renewable alternatives for horticultural growing media. Therefore, the objective of this study was to evaluate the potential of harvested …

Modeling And Mitigating Stale Hardcoded Secrets In Version Control History, Katarina Valentine Capalbo

USF Tampa Graduate Theses and Dissertations

Hardcoded secret vulnerabilities remain a growing and persistent problem that can be difficult to mitigate once exposed. While there has been significant research and advancements in identifying and preventing hardcoded secrets, less attention has been given to understanding the risks of leaving these secrets behind in version control history, as well as analyzing the sanitization of hardcoded secrets. These challenges present a growing gap between the risk and ability to mitigate and sanitize existing secret exposures. This thesis begins to address this gap by formalizing a threat model for hardcoded secrets that persist in Git version control history after the …

Kernelizing Protein Interaction Languages: Spectral Approximations And Random Fourier Features, Aishik Ghosh

Master’s Dissertations

Protein-peptide interactions play an important role in many biological phenomena, spanning adaptive immunity to disease pathology. In the Sliding Window Interaction Grammar (SWING) framework, interactions are represented as sequences of biochemical tokens embedded using Doc2Vec, allowing robust generalisation to unobserved MHC alleles. However, classification remains limited to a single Euclidean feature space that is incapable of resolving binding landscapes. This dissertation develops SWING for four distinct kernel types: Gaussian, Laplacian, anisotropic (ARD), and the Spectral Mixture (SM) kernel, each approximated using scalable Random Fourier Features. The SM kernel incorporates prior knowledge about secondary structure into its spectral density as biological …

Learning In Infants Using Intrinsically Motivated Goal Conditioned Reinforcement Learning, T I Darsan

Master’s Dissertations

Traditional artificial intelligence models learn by passively digesting large datasets. In contrast, human infants discover skills by actively interacting with their bodies and environments without explicit external rewards. This thesis introduces the Composer Architecture, a machine learning framework designed to mimic this autonomous, open-ended development. The Composer architecture operates in a multi-stage loop, the latent model using Contrastive Learning Through Time (CLTT) to compress high-dimensional raw data from visual, proprioceptive, and touch sensors into a low-dimensional space. To preserve data relationships and prevent topological collapse, a Softmax activation forces these latent representations to lie smoothly on a probability simplex. A …

Design And Evaluation Of A Code-Switching-Aware Multilingual Conversational Ai System Using Advanced Rag Architectures, Ashutosh Juvale

Master’s Dissertations

Conversational artificial intelligence has become the primary interface through which hundreds of millions of users in India seek information and customer support. Yet the way these users actually write and speak is fundamentally at odds with the monolingual assumptions baked into most retrieval and generation systems: they code-switch, fluidly mixing one or more of the twenty-two scheduled languages of India with English, frequently typing Indic words in the Roman script ("mera refund kab tak aayega"). Standard Retrieval-Augmented Generation (RAG) pipelines silently fail on such input — the retriever returns off-topic passages because the query and the knowledge base live in …

An Efficient Hierarchical Deployment Of Sensors For K-Coverage In Planner Wireless Sensor Network, Abhay Raj Singh

Master’s Dissertations

Ensuring reliable sensing coverage is a fundamental challenge in wireless sensor networks (WSNs), particularly when multiple sensors are monitoring each location to provide robustness against node failures. In this work, we address the problem of deterministic k-coverage in planar WSN by proposing a hierarchical triangular lattice-based deployment strategy that organizes sensor locations across di↵erent refinement levels and guarantees coverage of every point in the sensing domain by at least k sensors. Each lattice is three-colorable, and selective activation of color classes ensures adjustable coverage guarantees. We prove that activating a single color class at refinement level t guarantees at least …

Telluric Correction Of M-Dwarf Stars Using Machine Learning, Sayak Rana

Master’s Dissertations

The study of M-dwarf stars is of prime scientific interest to us because of their closer habitable zones and the favorable conditions they offer for exoplanet detection. However, telluric contamination of the ground-based spectra results in sharp absorption lines, which makes their study cumbersome. Removing this contamination is necessary for estimating key stellar parameters. The central contribution is a one-dimensional Convolutional Neural Network (CNN) that retrieves the four atmospheric parameters governing telluric absorption: pressure, temperature, humidity, and airmass. These predicted parameters are passed to Telfit which produces an estimated telluric spectrum. The observed spectrum is then divided by this estimated …

Morphodynamic Evolution Of Midnight Pass, West-Central Florida, Sydney C. Scott

USF Tampa Graduate Theses and Dissertations

This study takes advantage of a rare opportunity provided by the reopening of Midnight Pass due to two consecutive hurricanes in 2024, to examine the morphologic evolution of a newly reopened inlet and subsequent beach-inlet interactions. Historical aerial photos since the 1940s were collected and analyzed to understand the last closure of Midnight Pass, and to gain knowledge on the future of the newly reopened inlet. Time-series topography and bathymetry survey of the beach-inlet system were conducted bimonthly to quarterly for the first year and a half after the reopening, with the goals of quantifying the morphologic changes and understanding …

Efficiency Improvement Of Rag Based Slm For Edge Devices, Pavan Prashanth Avanigadda

Master’s Dissertations

The increasing need to deploy language models on constrained devices has given rise to efficiency issues in retrieval-augmented generation (RAG) approaches. Although RAGs boost answers’ quality by retrieving knowledge from external sources, current methods utilize static retrieval mechanisms, resulting in unnecessary computation, higher latencies, and inefficiency in resource usage. In this work, an efficient RAG approach based on small language models (SLMs) is presented, which uses a efficient and adaptive retrieval scheme. This method dynamically changes the retrieval depth and context constrution based on the complexity of the query, using a trained MLP router whose routing decisions are learned from …

Developing A Model To Generate More Digital Data Of Indian Languages For Multilingual Applications, Arya Bagde

Master’s Dissertations

Most of India’s scheduled languages remain critically under-served by language technology because parallel (translated) text — the raw material that modern multilingual systems depend on — is extremely scarce. Back-translation can synthesise such data automatically, but its quality varies enormously, and unfiltered synthetic data can be worse than no data at all. This dissertation develops a framework that generates synthetic parallel data for four low-resource Indian languages spanning three language families and four scripts — Assamese (Indo-Aryan, Bengali script), Bodo (Tibeto-Burman, Devanagari), Manipuri (Tibeto-Burman, Bengali script) and Santali (Austroasiatic, Ol Chiki)—and introduces CASCADE, a learned multi-signal quality gate that scores …

Dynamic Property Ordering For Efficient Multi-Property Bounded Model Checking, Vivek Kumar

Master’s Dissertations

Formal verification plays a critical role in ensuring the correctness of modern hardware designs. As the complexity of digital systems increases, designs are often associated with a large number of verification properties that must be analyzed within limited computational resources. In conventional multi-property bounded model checking (BMC), all properties are verified simultaneously. While this approach enables parallel analysis, difficult properties can consume a disproportionate amount of resources, causing simpler properties to be delayed and reducing the overall efficiency of bug detection. This thesis presents dynamic property ordering techniques for efficient multi-property verification using SAT-based bounded model checking in the ABC …

Adding 3,4-Methylenedioxy-Α-Cyclohexylaminopropiophenone To A Standard Operating Procedure (Sop) For Identifying Synthetic Cathinones Using Gas Chromatography-Mass Spectrometry, Napassawon Timpratoom

ETDs from 2020-2029

Synthetic cathinones were identified as new psychoactive substances (NPS) in the early warning advisory system by the United Nations Office on Drugs and Crime (UNODC). Synthetic cathinones have been commonly referred to as bath salts, which offered “legal highs” because they were not explicitly scheduled. These compounds were detected dur-ing the screening process in casework and required internal validation for confirmation. 3,4-methylenedioxy-α-cyclohexylaminopropiophenone was the target analyte for this study. It was alternatively named N-cyclohexyl methylone and “cyputylone." The aim of the study was to validate 3,4-methylenedioxy-α-cyclohexylaminopropiophenone into a Standard Operating Procedure (SOP) for identifying synthetic cathinones using gas chromatography/mass spectrometry …

Multi-Frequency Associative Memory For Continual Graph Learning Through Nested Optimization, Shuvam Kundu

Master’s Dissertations

Graph Neural Networks struggle to learn new tasks without forgetting old ones a problem known as catastrophic forgetting. In graph domains, this is compounded by structural shift, where newly added edges corrupt the learned representations of historical nodes even when model weights remain unchanged. We present CAM-Titans, a continual graph learning framework built around a two-buffer associative memory to address both parametric and structural forgetting. Our architecture operates across three timescales of adaptation: a slow base memory updated via ordinary gradient descent, an intermediate task buffer re-encoded after every task using the delta-rule, and a transient in-context state for rapid …

A Study Of Prompt Tuning On Small Language Models(Slms): A Controlled Benchmark And A Lightweight Instance-Aware Method, Narkadamilli Sahith

Master’s Dissertations

Parameter-efficient fine-tuning (PEFT) adapts a frozen pre-trained language model by training only a small number of additional parameters. Among PEFT approaches, prompt tuning prepends trainable continuous vectors (soft prompts) to the input. A recurring finding in the literature is that prompt tuning is strongly scale dependent: it rivals full fine-tuning on very large models but lags on smaller ones. This dissertation studies prompt tuning specifically in the small-language-model (SLM) regime. We (i) re-implement a representative set of prompt-tuning methods—Prompt Tuning, P-Tuning v2, LoPT, DPT, DePT, ACCEPT, Residual Prompt Tuning, and PARA—within a single controlled harness, enabling a fair head-to-head comparison …

Reproducing And Analyzing The “Lost In The Middle” And “The Power Of Noise” Phenomenon In Retrieval-Augmented Generation, Kousik Samanta

Master’s Dissertations

Retrieval-Augmented Generation has become the way to improve Large Language Models. They help with problems like knowledge and hallucinations. Recent studies show that these models still have limitations. One big problem is the “Lost in the Middle” phenomenon. Models can’t access information in the middle of contexts properly. Another counterintuitive observation is the “Power of Noise” paradigm, which suggests adding unrelated documents can actually make the generation better. We know these happen in extractive QA tasks, but we don’t know if they happen in tasks that need complex reasoning. This dissertation looks into how position and noise affect Long-Form Question …

Learning To See Lesions, Not Skin Tone: Counterfactual Multimodal Learning For Fair, Trustworthy, And Text-Free Dermatology Ai, Shivam Jangid

Master’s Dissertations

Recent advances in deep learning have significantly improved the performance of automated skin lesion classification systems, enabling accurate detection of various dermatological conditions from medical images. Despite these achievements, concerns regarding fairness and generalization remain a major challenge for the deployment of such systems in real-world clinical settings. A key factor contributing to these challenges is the presence of bias in training datasets, particularly with respect to skin-tone representation. Most publicly available skin lesion datasets contain a disproportionate number of samples from individuals with lighter skin tones. As a result, deep learning models trained on these datasets often learn representations …

Adaptive Spectral Trust Gate For Physics- Constrained Operator Learning, Soham Chakraborty

Master’s Dissertations

Physics-informed machine learning improves the plausibility, data-efficiency and generalization of surrogate models by injecting prior physical knowledge into the learning process. The current approaches can be broadly divided into two main categories: soft constraints, which add a physics residual to the training loss but guarantee nothing at inference time, and hard constraints, which project the model output onto the constraint set exactly but apply the projection uniformly to every part of the signal — including parts that are dominated by noise, discretization error, or model mismatch, where the idealized physics is not actually trustworthy. This dissertation proposes the Adaptive Spectral …

Deep Reinforcement Learning With Directed Asymmetry And Kolmogorov-Arnold Networks For Dismantling Interdependent Multiplex Networks, Soumyajit Dev

Master’s Dissertations

Identifying the minimum-cost node-removal sequence that fragments a complex network - the network dismantling problem is NP-hard and central to infrastructure resilience. In interdependent multiplex networks, this difficulty is compounded by cascading cross-layer failures. While deep reinforcement learning (DRL) agents utilizing graph neural network (GNN) encoders achieve near-optimal dismantling, current state-of-the-art architectures suffer from two critical limitations. Topologically, existing agents strictly assume undirected edges, rendering them inapplicable to directed systems - such as supply chains or gene regulatory cascades - where failure propagation is fundamentally asymmetric. To resolve this, we propose Disassembling Directed Interdependent Networks (DDIN). DDIN introduces an asymmetric …

New Studies In Lattice-Based Cryptography, Quantum Algorithms, And Privacy-Preserving Computation, Hansraj Jangir

Electronic Theses and Dissertations 2020 - Present

Classical public-key cryptographic schemes primarily rely on the presumed hardness of the integer factorization and discrete logarithm problems. However, Shor’s algorithm demonstrates that both problems can be solved efficiently on a sufficiently powerful quantum computer. This breakthrough motivated the search for quantum-resistant hard assumptions. The assumptions based on lattices are one of the primary candidates in post-quantum cryptography due to their strong theoretical foundations and well-established hardness against both classical and quantum attacks. In this thesis, we explore several directions of lattice based cryptology and quantum algorithms.

On the construction side, we propose two compact encryption schemes based on the …

Spectral Unmixing Using Machine Learning, Debashis Dhar

Master’s Dissertations

Spectral Unmixing is an important field of study nowadays which focuses on gener ating fractional abundance of each pixel into constituent materials .In this thesis we have tried to unmix each pixel into three end members namely glacial lake,debris and others with primarily focusing on glacial lake.We have performed various meth ods of linear spectral unmixing and non linear spectral unmixing. These methods are applied on the collected LandSat Data of east Himalayan terrain .Experimental results demonstrate the effectiveness of the proposed approach in achieving high accuracy and efficiency in glacier lake tracking on LandSat data.

A Switch-Point-Aware Contrastive Approach To Sentiment Analysis Of Hinglish Code-Mixed Text, Prasant Kumar Sahoo

Master’s Dissertations

With the increasing use of social media in non-English-speaking regions, especially in India, people often use Romanized Hindi and English together in their online communication. In a single sentence, they frequently mix Romanized Hindi and English, creating code-mixed text. However, most multilingual transformer models are pre-trained primarily on monolingual data. As a result, NLP systems face challenges when processing code-mixed text, as a single word may be fragmented into meaningless subword pieces, making it difficult for the model to capture its semantic meaning accurately. In this dissertation, we propose a parameter efficient neural architecture consisting of three main components to …

Magnetic Field Evolution In Galaxy Clusters, Pranaav S

University Departments

In this work, we investigate the magnetic field evolution of 352 massive galaxy clusters from the TNG-Cluster cosmological magnetohydrodynamic simulation. Clusters are classified ac-cording to their dynamieal state using a combination of Merger 'Weight parameter (o') and central entropy (

cosmic time and examine the spatial distribution of magnetic energy. Our results reveal a pos-itive correlation between cluster mass and magnetic field strength, indicating higher magnetic amplification in massive clusters. In our stuffy, magnetic field undergoes rapid growth during the early stages of cluster formation (from z = 10 to z = 2), fo11owe‹1 by a gradual transition into a …

Lepton Flavor Violating Z And Higgs Boson Decays In Left-Right Models, Mangala Hegde Nagapati

University Departments

In this work, we explore lepton flavour violating (LFV) decays of the Z and Higgs boson in Alternative Left-Right Model that extend the Standard Model field content and gauge structure. The presence of exotic fermions, additional gauge bosons, and charged scalar fields in these scenarios can induce LFV processes through I-loop and mixing effects. We evaluate the 1-loop branching ratios for decays such as Z -t et,e1and h -t etej (i -=t- j), taking into account current experimental constraints on exotic field masses, and flavour-violating observables. A random scan of the model parameters is performed to systematically examine the phenomenology …

Formation Pathways Of Complex Organic Molecules In The Interstellar Medium- A Computational Study, Sharadhi Vakuda

University Departments

Nitrile-substitutedpolycyclic aromatic hydrocarbons (CN-PAHs) are of growing astrochemical interest because they connect aromatic carbon chemistry with the rich nitrile chemistry observed in interstellar environments. Motivated by the recent detection of cyanocoronene (C24H11CN) in Taurus Molecular Cloud- I (TMC-1), this thesis investigates both the possible formation pathways of cyanocoronene and its infrared detectability in astronomical environments. In the first part of the work, several chemically plausible reaction routes were explored using density functional theory (DFT), along with higher-level multireference calculations including CASSCF and NEVPT2 for selected key structures. Geometry optimizations, vibrational frequency calculations, and transition-state searches were carried out to construct …

(Re)Claiming Homeplace: Dance As Embodied Resistance & Healing For Health & Spatial Justice, Preeya G. Kannan

University Honors Theses

This thesis argues that dance functions as both an upstream public health intervention and a form of radical spatial resistance; creating healing, agency, and belonging among historically marginalized communities. Drawing on bell hooks' concept of homeplace, Jasbir Puar's assemblage theory, Nancy Krieger's eco-social theory, and Katherine McKittrick's Black geographies, I position the body as both an ecological and political landscape shaped by histories of colonialism, racial capitalism, displacement, and resilience.

Through this community-based participatory research study, Homeplace, I explore dance as a form of social prescription that centers liberation rather than pathology. Movement is often reduced in public health discourse …

Extending Unibreak: Semantic Retrieval And Harmful-Intent Direction Suppression For Token-Level Llm Jailbreaking, Sanket Saha

Master’s Dissertations

Token-level adversarial perturbations remain one of the most efficient known attacks against the safety alignment of instruction-tuned large language models (LLMs). Among recent works, the UniBreak framework (You et al., 2026) stands out for unifying gradient-based optimization with an evolutionary perturbation repository. However, its repository relies solely on accumulated success frequency without utilizing query content, and its fitness function implicitly assumes that suppressing refusal tokens is sufficient to elicit harmful responses. In this dissertation, we extend UniBreak along both axes and re-evaluates the framework under stricter generalization and judgment protocols. Specifically, we introduce a semantic perturbation repository that replaces frequency-only …