Digital Commons Network^™

Moral: Multimodal Reasoning For Autonomous Language Models With Sensor-Grounded Spatial Bev Rendering, Ambarish Govindarajulu Kaliamurthi

Master's Theses

Autonomous-driving vision-language models describe scenes fluently but reason poorly about metric, safety-critical spatial relationships because they do not read sensor geometry in a grounded way. This thesis presents MoRAL (Multimodal Reasoning for Autonomous Language Models), a two-stage fine-tuning pipeline that teaches a compact 2-billion-parameter VLM to decode a physics-encoded Bird’s Eye View (BEV) representation – LiDAR distance as color, object class as cluster shape, radar Doppler velocity as directional wedges – and then trains it to reason over that representation for driving decisions. Stage 2 then fine-tunes on 57,696 teacher-generated chain-of-thought examples across eight question types, using Cosmos-Reason2-8B as teacher …

Effects Of Variable Light On Sexual Competition In Kelp Species, Jaycee Lanza

Master's Theses

As foundation species, kelps significantly influence the structure and composition of their surrounding habitat by modifying environmental conditions, including light levels, currents, nutrient availability, sedimentation, and seawater chemistry. Light, in particular, is a critical resource that varies greatly within kelp forests, affecting kelp reproductive processes such as gametogenesis, fertilization, and recruitment, which are vital to population persistence. Biological factors, such as density-dependent competition for resources and sexual competition with other kelps, can also affect such processes. To understand how environmental variables influence kelp populations, it is crucial to recognize their combined effects. This study investigated the impacts of light levels …

The Zeal Instruction Set Architecture, Joseph A. Gerani

Master's Theses

The Instruction Set Architecture of a CPU (Central Processing Unit) determines what type of instructions the CPU is able to understand, how those instructions are encoded, and what it should output upon receiving those instructions as input. There are currently three popular ISAs meant for the consumer market: x86, RISC-V, and ARM, as well as a fourth that mostly now exists in the server market by the name of Power. One of the most important parts of an ISA is for engineers to be able to understand it and make use of it. If an ISA is too complicated, nobody …

Error-Tolerant Metric Dimension, Leander Ten Hoff

Master's Theses

Metric dimension is a graph parameter that measures the smallest distance-based unique coordinate system on a graph. Fault-tolerant metric dimension is a variant that requires redundancy. Inspired by this idea, we propose and investigate a new variant we call error-tolerant metric dimension. We first prove some fundamental results to gain familiarity with this more complex variant. Then, we use these tools to study relative behaviors of error-tolerant metric dimension. We prove explicit computations for simple graph families, including bipartite complete graphs and paths. Finally, we extend extremal results from previous papers on metric dimension, including a full correction of an …

Data Augmentation For Vision-Language-Action Models: Bridging Vision And Language, Miaosen Zhou

Master's Theses

This thesis focuses on real-time task execution and object detection for autonomous robots through dataset augmentation. We propose a data augmentation approach to address dataset imbalance in Vision-Language-Action (VLA) models across both image and text modalities during the fine-tuning process. The proposed method takes an image as input and generates a structured textual description using a prompt engineering strategy to augment the textual input. The generated augmented text includes key elements such as the task goal, scene description, reasoning, and execution plan, along with other relevant contextual information. This enriched representation improves the quality of the training data and supports …

Towards Neural Network Optimization: Addressing Issues With Corrupted Weights Within Models, Nick Najafizadeh

Master's Theses

Neural networks are a recent popular technology inspired from human brains. Much of their popularity arises from how they excel in reasoning and logic, and are generally rather efficient in their tasks. With those strengths, they are frequently used in transportation and business among many other fields. However, neural networks have many factors that can deteriorate their performance, one of the most critical being weight corruption. Therefore, it is of utmost importance to detect and handle them as soon as possible so as to minimize the negative impact on a network’s performance. The optimization of neural networks would be especially …

Accessibility Fairness Practices In Ai Applications For People With Disabilities, Megan Gross

Master's Theses

Advancements in artificial intelligence (AI) improve the lives of people every day with tools like the auto-captioning of videos, improved screen-reader capabilities, and advanced mobility control through speech. However, are all groups of people benefiting from AI or are some being overlooked and left out? Although AI tools made for people with disabilities (PWDs) have improved their lives, AI for the general population generally ignores the experiences of PWDs, making them unable to interact with and benefit from technology. This research evaluates ChatGPT and Gemini in Gmail for usability fairness and analyzes how current regulations and development processes fail to …

Diagrammatic Methods For Calculating Correlation Functions In Low-Dimensional Quantum Gravity, Nicholas A. Parrilla

Master's Theses

This thesis investigates connections between random matrix theory and low-dimensional quantum gravity through the spectral form factor (SFF), a well-studied function that characterizes correlations in the eigenvalue spectrum of an operator. In random matrix ensembles, a feature of the SFF known as the “ramp” is governed by universal spectral correlations, and in Jackiw–Teitelboim (JT) gravity this regime is due to the double-trumpet wormhole geometry. We study the connected correlation functions underlying the SFF using combinatorial, diagrammatic methods from a mathematical formalism known as second-order free probability theory, where they are expressed in terms of annular noncrossing permutations. To access a …

Rewriting The Calculus 3 Workshop At San José State University, Moorea Lippert

Master's Theses

Calculus 3 at San José State University is a very important course for any student pursuing a STEM degree. However, the course is very challenging, so the university offers a supplemental workshop course for Calculus 3 students. Unfortunately, this workshop has not aligned with modern research in math education. This thesis documents efforts to recreate the workshop to better align with this modern research.

Hydrus Modeling Of A Sahelian Agroforestry System: An Investigation Of Local Water Balance, Nicole A. Butler

Master's Theses

Effects of drought can devastate agricultural productivity in the West African Sahel. Climate complexities, such as lower-than-average rainfall and high winds, can lead to crop failure and low yields, thereby creating food security issues. Agriculture in the Sahel is primarily rainfed, and the rain is becoming increasingly unpredictable due to the changing climate. The method of intercropping groundnut (Arachis hypogea) and pearl millet (Pennisetum glaucum) with the shrub species Piliostigma reticulatum can combat the effects of climate change and restore degraded soil. Intercropping can increase water availability and crop yield when compared to crops grown in the absence of shrubs. …

Pipeline Optimization Of Agentic Llms For Text-To-Sql, Peter Conant

Master's Theses

Current database interfaces limit users’ interaction to those with technical skills, creating timely roadblocks for non-technical professionals. Text-to-SQL aims to simplify database interactions by translating natural language questions into database queries, but long-standing challenges like question understanding, question-schema linking, and SQL generation have held the field back. In the AI era, foundational LLMs prove to be very capable of question understanding SQL, perform well in Schema Linking, Generation, and Evaluation tasks. However the cost to run these model is a hurdle for many organization with low funds and resources. Text-to-SQL solutions often operate across large enterprise size databases with, and …

Monoids With K-Strictly Local Geodesic Languages, William M. Hong

Master's Theses

A paper by Gilman, Hermiller, Holt and Rees (2011) prove that a group G finitely generated by X is virtually free if and only if the language of geodesics words in G over X is k-strictly local if and only if the word problem is context-free. If M is a monoid finitely generated by X, we define Γu(M,X) to be the underlying, undirected graph of Γ(M,X). We prove that if the language of geodesics in Γu(M,X) based at the identity is k-strictly local, then the monoid and semi-group word problems are context-free.

A Closer Look At The Relationship Between Workplace Harassment And Negative Workplace Behavior: The Mediating Role Of Revenge Cognitions And The Moderating Role Of Fear Of Internal Threat, Keely L. Sinclair Smith

Master's Theses

The purpose of the current study was to provide a closer look at the relationship between workplace harassment and negative workplace behavior by investigating the mediating role of revenge cognitions and the moderating role of fear of internal threat, a novel construct that included both formal and informal organizational sanctions. A sample of 103 employees from a variety of occupational backgrounds participated in the study by completing an online survey. Results of the simple mediation analysis showed that workplace harassment was positively related to negative workplace behavior, and revenge cognitions fully mediated this relationship. In other words, employees who experienced …

Perceptions Of Speech-Language Therapy Among Generations Of One Mexican-American Family, Sabrina A. Macias

Master's Theses

This research is a modified replication of Salas-Provance et al. (2002) and analyzes the perceptions of speech-language therapy among one Mexican-American family. It investigates opinions about causes of speech-language disorders, for whom speech therapy can be helpful, characteristics and stereotypes about people with hearing and speech disorders, and the influence of education level on such opinions. With the assistance of three family members and two cultural informants, an online survey was developed consisting of demographic factors, Mexican folk and medical beliefs, and potential stereotypes of people with speech, language, and hearing disabilities. Twenty-three family members (response rate = 76.7%) completed …

Extracellular Vesicle-Mediated Transfer Of Gsdme-Nt Pores As A Mechanism Of Inflammation Propagation, Bifeng Xiao

Master's Theses

Inflammasomes are multiprotein complexes that initiate inflammatory signaling through activation of caspase-1, cytokine maturation, and cleavage of Gasdermin D (GSDMD). Cleaved GSDMD forms membrane pores that contribute to cytokine release and pyroptotic cell death. Recent evidence suggests that extracellular vesicles (EVs) may facilitate the intercellular transfer of inflammatory mediators. This study investigated whether GSDMD N-terminal fragment (GSDMD-NT) associates with EV-enriched fractions and contributes to inflammation propagation. Doxycycline-inducible HeLa cells expressing FLAG-tagged GSDMD-NT were used to generate EV-enriched fractions. Western blot analysis detected GSDMD-NT-associated material within EV fractions, while EV marker analysis supported successful enrichment of EV-associated material. Exposure of recipient …

Dissecting The Discourse: Colin Kaepernick's National Anthem Protest And Conservative Media As Competing Racial Projects, Travis Moody

Master's Theses

This thesis examines the media portrayal of Colin Kaepernick’s protest against police brutality and racial injustice during the National Anthem in the 2016 National Football League (NFL) season. Using content analysis, I analyze why Colin Kaepernick protested in his own words and how conservative media outlets such as Fox News rearticulated and criticized his social justice stance as an unpatriotic denigration of the American flag and Star-Spangled Banner. As I further argue, social movements such as #BlackLivesMatter and the Colin Kaepernick-inspired “Take a Knee” movement represent a political project to inform racial discourse regarding racial injustice in the United States. …

Evaluation Of A Sleep Health Composite: Changes In Sleep Over The Menarcheal Transition Among Girls, Julianna Adornetti

Master's Theses

Sleep undergoes a shift during puberty, increasing the risk for psychological and insomnia symptoms, particularly among girls. However, little research has examined how the onset of menarche may contribute to vulnerability in sleep and psychological functioning. The Sleep Health framework provides a standardized, multidimensional way to conceptualize sleep health (RU-SATED: satisfaction, alertness, timing, efficiency, duration, and regularity). Although prior studies have provided proof-of-concept for assessing sleep health using a composite score (SHc), its measurement properties have yet to be explored. The primary aims of this longitudinal study were to (1) evaluate the factor structure of a sleep health composite, (2) …

Growing Through Systemic Adversity: Evaluation Of Developmental Trauma Assessment Among Black And/Or Latiné Adolescents With Adhd, Victoria Grant O'Daniel

Master's Theses

Black and Latiné youth are disproportionately impacted by targeted harm and trauma exposure, which is further exacerbated by stigma, biases, and punitive actions towards youth with attention deficit/hyperactive disorder (ADHD). This presents an urgent public health concern, as frequent exposure to trauma across development can lead to difficulties with behavior regulation, interpersonal relationships, impulse control, emotion regulation, and attention and cognition. Developmental trauma disorder (DTD) was proposed to capture the adverse developmental outcomes experienced by youth exposed to chronic trauma; however, there is a dearth of research on how to assess DTD. Thus, the current study aimed to establish support …

Trim-Transfer: A Transfer Learning Approach For Cross-Trim Level Can Intrusion Detection, Baylor J. Whitehead

Master's Theses

Modern vehicles contain many Electronic Control Units (ECUs) that communicate through CAN. While CAN enables efficient data communication, it lacks built in authentication and encryption, allowing adversarial actors to inject malicious CAN messages. This limitation has motivated the development of CAN intrusion detection systems (IDS). However, deploying IDS across a vehicle lineup requires collecting large labeled datasets and retraining models, increasing development cost and limiting scalability.

This thesis investigates the use of transfer learning with LSTM-based deep neural networks to reduce retraining cost while maintaining model detection performance. A baseline LSTM model is trained using CAN data from a base …

Exploring Regenerative Vegetable Systems On The Central Coast Of California, Una S. O'Connell

Master's Theses

As climate change and current agricultural practices place pressure on the long term sustainability and resilience of food production systems, there is growing interest in whether existing agricultural systems can adapt to maintain productivity while improving environmental outcomes.

This study evaluated the effects of regenerative versus standard organic vegetable production systems on yield, insect communities, and weed suppression across three field trials conducted on California's Central Coast. In addition, I collected baseline soil health data for all trials. Trials 1 and 2 were carried out at the Cal Poly San Luis Obispo Organic Farm using cabbage and broccoli as cash …

A Comparative Study Of Model Predictive Control And The Stanley Method For Vehicle Path Tracking Applications, Noah S. Fitzgerald

Master's Theses

This thesis compares a model predictive controller (MPC) and a lateral Stanley controller for vehicle path-tracking applications under simulation-based and perception-driven operating conditions. Both controllers were evaluated in simulation using a nonlinear dynamic bicycle model executing single and double lane change maneuvers. Following simulation-based evaluation, both controllers were implemented on hardware within a perception-driven steering-control pipeline. This pipeline utilized recorded sensor data from the MXcarkit 1/8^th-scale autonomous vehicle platform, incorporating lane instance segmentation and homography-based roadway estimation.

Under idealized simulation conditions, the MPC demonstrated improved trajectory-tracking performance during aggressive maneuvers while requiring greater steering activity and computational effort …

Developing A Humpback Whale Vocalization Detector Using Machine Learning Models, Lucas Kantorowski

Master's Theses

Humpback whale songs are notoriously complex. Identification of humpback whale song units requires bioacousticians to tediously listen, analyze, and annotate collected sound data. Even sparse data requires listening to the entirety of the collected acoustic data. In this study, three hours of audio containing over one-thousand humpback whale song units was collected in Monterey Bay, California.

Prior studies have seen success using convolutional neural networks by performing image classification on hundreds of hours worth of spectrograms. Our study uses traditional machine learning models, as they are less computationally demanding, and require less data.

We use time splitting and Mel-frequency cepstrum …

Discovering Differentially Correlated Sets Across Multiple Treatments, Tyler J. Stoen

Master's Theses

When comparing data sampled under several different conditions, we often want to study not only the overall treatment effect but also how treatments influence the relationships among variables. Differential Correlation Mining (DCM) is an existing method for identifying groups of variables whose average pairwise correlation differs significantly between exactly two sampling conditions. We extend this framework to the setting of three or more treatments, where the goal is to identify variable sets whose pairwise correlation structure varies across all conditions simultaneously. Our proposed method, VSAT for three or more treatments, generalizes the DCM search procedure using a multi-treatment hypothesis test …

How Human Vs. Ai Chatbots Support Psychological Needs Of Students, Tymon T. Vu

Master's Theses

AI auto graders and virtual assistants are increasingly prevalent in educational set- tings. Although current literature demonstrates that AI chatbots can positively im- pact student motivation, there is little definitive evidence exploring whether these systems can replicate the effective motivational methods used by human teachers. Thus, in this experiment, we investigate how the perception of the identity of a tutor (AI versus human) impacts the psychological needs of students. 27 undergraduates taking introductory computer science courses were asked to explain code in plain English (EiPE). Participants received feedback from two tutors in within-subjects fashion: one human and one labeled as …

A Study On Quantization And Hardware Cost Tradeoffs In A 5g Nr Ldpc Decoder On Fpga, Giancarlo Acosta

Master's Theses

The fixed-point implementation of a 5G New Radio LDPC decoder forces a tradeoff between precision and hardware cost, governed by the variable node word length W_L, the check node message width W_R, and the normalized min-sum correction factor α. This thesis characterizes how these three parameters affect both error correction performance and FPGA resource utilization for an LDPC decoder on an established LDPC decoder architecture. A bit-accurate MATLAB core model records bit error rate (BER) and frame error rate (FER) while a verified HDL Coder model generates synthesizable VHDL for Vivado synthesis, and both are swept …

Movement And Space Use Of Weaned Northern Elephant Seal (Mirounga Angustirostris) Pups During Their First At-Sea Foraging Migration, Katie Saenger

Master's Theses

The northern elephant seal (Mirounga angustirostris) is a marine mammal known for engaging in long foraging migrations at sea. Adult male northern elephant seals follow pathways along the western coast of North America towards Alaska and adult female northern elephant seals follow pathways to open ocean in the North Pacific. However, there is little information on weaned northern elephant seal pup foraging behavior during their first foraging migration. This lack of knowledge is significant because northern elephant seals have a mortality rate of >50% in the first year. This project aimed to better understand the at-sea migratory movement …

Reproducing And Evaluating Charger Surfing: Robustness Of Smartphone Charging-Line Side-Channels, Colby M. Watts

Master's Theses

Smartphones are frequently connected to external, untrusted charging hardware, creating opportunities for side-channel attacks that do not require malware or direct access to device data. Charger Surfing, a recently proposed charging-line power analysis side-channel attack, reported high accuracy in inferring touchscreen input from voltage measurements collected from a smartphone’s charging cable; however, the reproducibility and robustness of these results under different conditions remain unclear. This thesis presents an independent replication and evaluation of Charger Surfing, including the development of an end-to-end data collection pipeline consisting of a modified charging cable, oscilloscope-based recordings, custom Android app, automated trace processing, and convolutional …

Preserving Structural Alias Information Across The Mlir-To-Llvm Lowering Boundary, Shravan Sheth

Master's Theses

Machine learning training and inference workloads run at massive scale, where the efficiency of generated machine code directly affects throughput and energy consumption. The compilers that lower model specifications to hardware instructions rely on optimization passes that can only exploit information visible in the representations they operate on. MLIR, the intermediate representation framework underlying production machine learning compilers such as IREE and Triton, is designed so that each abstraction level can encode semantics that lower levels cannot represent. One example is memref.subview, which partitions a buffer into typed regions with explicit offsets and sizes, making structural non-overlap provable from the …

Design And Analysis Of A Customizable 6-Dof Force/Torque Sensor, Pyeongkang Kim

Master's Theses

This thesis presents the design, development, calibration, and testing of a novel customizable six-degree-of-freedom force/torque sensor, called the Kustom Force/Torque Sensor (KFTS), for use in the Cal Poly Low Speed Wind Tunnel (LSWT). The KFTS was developed as an alternative to commercial multi-axis force/torque sensors that feature fixed specifications. The swappable components and configurable structure of the KFTS enables changes to the sensor sensitivity, sensing range, and frequency response. Experimental testing verified that the sensor behaved as a linear time-invariant system, enabling calibration through the method of least squares and allowing the applied forces and moments to be reconstructed as …

Praxsmith: An Actor-Centric Domain Specific Language For Emergent Narratives In Video Games, Jacob C. Kelleran

Master's Theses

The video game industry is rapidly growing, and interest in game development is growing alongside it. Among this growing interest is a desire to include branching and emergent narratives in games. This subject is understood to be difficult, as every possible trajectory within the story creates exponential complexity. Large studios are sometimes able to create convincing implementations, but that is often with hundreds of developers and writers. Over the years, various solutions have been proposed to create frameworks and languages that aim to tackle this problem, with varying degrees of success. Some frameworks attempt this by placing the focus on …