Physical Sciences and Mathematics Commons^™

Ai-Driven Personalized Radiotherapy Planning, Nithin Venkatesh, Marco Pota, Maged Shaban

SAML-25 Workshop on Statistical and Machine Learning

The planning of radiation oncology treatment is made more dynamic and individualized by Artificial Intelligence (AI). Routine radiotherapy practice applies normative procedures indifferent to patient-specific parameters such as tumor volume, patient anatomy, and heterogeneity in the delineation of treatment response. Inadequate and over-radiation treatment is the most prevalent outcome. Further, with the inclusion of AI, it can facilitate enhancing the healthcare industry through optimizing radiotherapy using an array of patient information such as molecular profiles and imaging data. The product offers an end-to-end AI-driven solution to all aspects of radiotherapy, from initial consultation (diagnosis) to adaptive treatment planning. All the …

Early Lane Change Prediction For Mixed Traffic With V2x Communication, Muhammed Fatih Koc, Nouman Ashraf, Pramod Pathak, Sachin Sharma

SAML-25 Workshop on Statistical and Machine Learning

Lane change prediction is essential for ensuring road safety and effective decision-making in autonomous vehicles (AVs). AVs will probably take several decades to penetrate new vehicle sales. As AVs and human-driven vehicles (HDVs) will coexist in traffic for the long term, AVs must understand the lane change intentions of surrounding HDVs. Lane changing is a critical manoeuvre that can cause a crash if it is performed late or if incorrect lane adjustments are made. Therefore, forecasting surrounding vehicles’ lane change intentions in advance is essential to ensure safe driving in mixed traffic environments having both AVs and HDVs. The unpredictability …

A Machine Learning Approach To Improve Prediction In Chemical Exposure Risk Assessment, Michele Marro, Cédric Koller, Hasnaa Chettou, David Vernez

SAML-25 Workshop on Statistical and Machine Learning

Exposure models play a crucial role in predicting chemical exposure in workplaces, offering an essential alternative to measurements, which are resource-intensive and time-consuming and sometimes not possible. Despite their widespread use and continuous development, significant challenges persist, including variability in predictions, limited model updates, and difficulties in accessing the required input data. In this study, we investigate how modern machine learning techniques can contribute to the improvement of exposure models by addressing these limitations. To overcome the frequent lack of data, we explore the use of synthetic datasets generated through existing exposure models. This approach allows for the study of …

Interpretable Ai In Education: A Comparison Of Glass-Box Models For Predicting Student Success, Jan Glazenborg

SAML-25 Workshop on Statistical and Machine Learning

This Master’s thesis addresses early identification of first-year Computer Science students at risk of underperformance by comparing inherently interpretable (“glass-box”) predictive models with the existing Naïve Bayes–based PreSS tool. The PreSS dataset was originally compiled by Quille & Bergin from 692 first-year CS1 students across eleven institutions in Ireland and Denmark, who completed surveys on programming and mathematics backgrounds, gaming habits and a short programming test four to six hours into the course. Seventeen normalized features capturing demographic, academic and behavioural factors were extracted. In this thesis, four machine learning models are evaluated: Naïve Bayes, explainable boosting machines, automatic piecewise …

Pros & Cons Of Reinforcement Learning - Illustrated By The Problem Of Controlling Gantry Robots, Horst Zisgen

SAML-25 Workshop on Statistical and Machine Learning

In this talk a solution for the dynamic scheduling of flexible flow shop systems using gantry robots for material handling by means of simulation and Reinforcement Learning (RL) is presented. Subsequently the pros and cons of a RL approach are briefly discussed and illustrated at the robot control problem.

Survival Predictions From Classification Algorithms – Concepts And Application To Graft And Patient Survival After Kidney Transplantation, Antje Jahn

SAML-25 Workshop on Statistical and Machine Learning

Clinical prediction models are developed to predict long-term patient outcomes following medical interventions. One example motivating this research is the prediction of graft and patient survival after kidney transplantation, using data from the German organ transplantation registry. A practical issue in this context is to deal with incomplete information due to right-censoring, which arises when patients are lost to follow-up or enter the study at different times, resulting in varying durations of observation. This is particularly relevant in the registry data, where follow-up is frequently incomplete or irregular. While traditional survival analysis methods handle censoring by modeling the hazard function, …

Pathology’S Place In Understanding The Bias And Inequalities In Women’S Healthcare, Andrea Heaney, Eugene Hickey, Emma Murphy

SAML-25 Workshop on Statistical and Machine Learning

Women’s healthcare is a complex, multifaceted issue with both historic and implicit biases, along with biological differences between men and women. With the advancement of AI tools in healthcare and the potential for biased data to create biased models, it is vital to consider how women are represented in data. Previously conducted semi-structured semantic interviews with clinicians were analysed via Braun and Clark’s method of thematic analysis. The analysis of these interviews yielded the following themes: Gender Influencing Health, Pregnancy, Social Factors, General Health, Treatment, Training, and Research. These themes highlight that context is key to understanding the biases in …

Impact Of Spatial Diversity And Subject Variability On Wifi-Based Human Activity Recognition, Amany Elkelany, Robert J. Ross, Susan Mckeever

SAML-25 Workshop on Statistical and Machine Learning

In recent years, WiFi-based Human Activity Recognition (HAR) has gained substantial attention due to the ubiquity of WiFi infrastructure and advancements in wireless communication. Unlike camera-based systems that raise privacy concerns or wearable sensors that require user compliance, WiFi-based HAR provides a noninvasive and practical alternative that operates seamlessly with existing infrastructure. WiFi-based HAR leverages fluctuations in wireless signals, particularly Channel State Information (CSI), to passively detect and classify human activities. WiFi-based HAR models often achieve high accuracy in a single environment but suffer significant performance drops when applied to new environments due to variations in spatial settings, human movement, …

Optimising Ai For Chemical Imaging: Benchmarking Performance Against Foundation Models For Task-Specific Applications In Histopathology, Rahul Suresh, Mohd Rifqi Rafsanjani, Karin Jirstrom, Arman Rahman, William M. Gallagher, Aidan Meade

SAML-25 Workshop on Statistical and Machine Learning

The integration of chemical imaging with artificial intelligence presents a compelling route toward fully digital, label-free histopathology, yet it also introduces notable challenges. While deep learning models from domains like machine vision, digital pathology, and remote sensing are readily accessible, they frequently struggle to generalize effectively to chemical imaging data, as highlighted in recent research [1]. Additionally, although foundational pathological models show potential for advancing AI-based histopathological diagnostics and prognostics, our preliminary assessments suggest they may fall short in addressing the broad spectrum of classification tasks encountered in clinical settings. In this presentation, we highlight some recent published work from …

Exploitation Of Plasma Treatment-Assisted Monocarboxylic Cobalt Phthalocyanine Nanorod Growth For High-Efficiency Photodetection Applications, Ahmed Ramzy, Ahmed M. El-Mahalawy, M A. Abd El Ghaffar, Wael A. Abbas

Nanotechnology Research Centre

Here, the effect of cobalt phthalocyanine (CoPC) functionalization by a monocarboxylic group in the form of thermally evaporated thin films followed by plasma treatment for variable time intervals is introduced. The effect of plasma treatment on crystalline structure, molecular structure, and morphology was examined using the XRD, FT-IR, and FE-EEM techniques. Significant modifications were observed in the crystal structure and morphology with increasing plasma treatment time intervals. The plasma treatment for 5 min enhanced the formation of the nanorods along the deposited surface with improved crystallinity. Tracking the optical properties variation of the deposited MCCoPc thin films under plasma treatment …

Automated Approaches To Interpreting And Explaining Machine Learning Models, Tilak Chandrashekar, Tarry Singh, Maged Shaban

SAML-25 Workshop on Statistical and Machine Learning

This paper proposes a novel framework for automating ML model interpretation and explainability across different applications with an emphasis on transparency, trust, and human-centric decisionmaking assistance. Although ML models, particularly advanced structures such as deep neural networks, possess superior predictive powers, their interpretability tends to be obscure, and thus their application in sensitive or regulated domains is impeded. Current XAI techniques, though promising, tend to be post-hoc and are not scalable for real-time or large-scale deployments. This study addresses such concerns by presenting an automated, modular pipeline where interpretation techniques are embedded in the process of developing the ML model. …

Tracking The Kinetics Cellular Glycolysis And Glutaminolysis Pathways Using Vibrational Spectroscopy, Combined With Multivariate Statistical And Machine Learning Approaches For Data Mining, Zohreh Mirveis, Nithin Patil, Hugh Byrne

SAML-25 Workshop on Statistical and Machine Learning

Understanding dynamic metabolic processes within living cells is crucial for gaining insights into cellular function and disease mechanisms. The kinetics of glycolysis and glutaminolysis pathways play significant roles, as alterations in their activity have been linked to various disorders, including cancer and mental health conditions such as bipolar disorder. These pathways therefore hold potential as biomarkers for disease diagnosis and therapy. However, real-time monitoring of their kinetics remains challenging due to the lack of suitable non-invasive techniques. Current gold-standard fluxomics approaches, such as mass spectrometry, are destructive to cells and thus unsuitable for time-resolved studies. In this study, we evaluate …

A Framework For Scalable And Controlled Hallucination Data Collection, Lin Ting Liang

Computer Science Senior Theses

This thesis addresses a key bottleneck in hallucination research: the scarcity and limitations of hallucination benchmark datasets. Existing datasets typically focus on a single type of hallucination and are expensive to produce due to the need for manual prompt creation and annotation. To overcome these challenges, we propose a novel mixture-of-experts (MoE) adversarial framework that actively induces hallucinations. Our framework employs three large language model (LLM) agents that iteratively and adversarially revise prompts to provoke hallucinated responses from a target question-answering model. It automates the generation of both intrinsic hallucinations (logical inconsistencies) and extrinsic hallucinations (inclusion of unverifiable external information). …

Spatial Analysis Of La Nana Bayou Watershed To Assess Stream Health, Rylee G. Gabbert

Electronic Theses and Dissertations

The purpose of this study was to evaluate the feasibility of using land cover mapping to identify water quality indicators within a river basin, assessing whether this method provides greater efficiency compared to traditional field-based water quality testing. Land cover mapping has efficiently monitored environmental changes by detecting alterations within specific areas. With the La Nana Bayou Watershed positioned in the heart of Nacogdoches City, an urbanized environment; it is subject to human induced alterations that can negatively affect the natural functionality of its system. When water quality indicators are successfully related to land cover maps, changes within the landscape …

Ensemble-Based Binding Free Energy Profiling And Network Analysis Of The Kras Interactions With Darpin Proteins Targeting Distinct Binding Sites: Revealing Molecular Determinants And Universal Architecture Of Regulatory Hotspots And Allosteric Binding, Mohammed Alshahrani, Vedant Parikh, Brandon Foley, Gennady M. Verkhivker

Mathematics, Physics, and Computer Science Faculty Articles and Research

KRAS is a pivotal oncoprotein that regulates cell proliferation and survival through interactions with downstream effectors such as RAF1. Despite significant advances in understanding KRAS biology, the structural and dynamic mechanisms of KRAS allostery remain poorly understood. In this study, we employ microsecond molecular dynamics simulations, mutational scanning, and binding free energy calculations together with dynamic network modeling to dissect how engineered DARPin proteins K27, K55, K13, and K19 engage KRAS through diverse molecular mechanisms ranging from effector mimicry to conformational restriction and allosteric modulation. Mutational scanning across all four DARPin systems identifies a core set of evolutionarily constrained residues …

Benchmarking Energy And Performance Of Parallel Machine Learning Models Using Hardware And Software Power Meters, Urooj Asgher, Tania Malik

SAML-25 Workshop on Statistical and Machine Learning

The growing reliance on machine learning algorithms across domains such as healthcare, transportation, and finance has led to their increased deployment on high-performance computing platforms. While performance optimization remains a central concern, energy efficiency is emerging as a critical design consideration, particularly in light of global sustainability goals. This study presents a comparative analysis of the energy consumption and performance of serial and parallel implementations of four machine learning algorithms, K-means clustering, Ant Colony Optimization, Logistic Regression, and Random Search. Experiments were conducted on an HPC testbed using both hardware-based and software-based power meters to measure energy consumption. The results …

Partitioning Around Medoids On Product Spaces: A Clustering Approach For Cylindrical Data, Yahia Hammami, Houyem Demni, Amor Messaoud, Giovanni C. Porzio

SAML-25 Workshop on Statistical and Machine Learning

Clustering is a common unsupervised task in data analysis and machine learning. It deals with finding clusters of objects that are characterized by the highest similarity within the same cluster and the highest dissimilarity between different clusters. One of the most used algorithms in clustering is the popular Partitioning Around Medoids (PAM), also known as k-medoids [4, 5]. The algorithm imposes the center of clusters to be some of the data points, and it looks for a minimal value of the sum of the dissimilarity to all the objects. One of the recognized properties of such a method is its …

Improving Node Classification For Graphs Withweak Feature Signals: A Similarity-Entropy Aggregation Approach, Brian Daniel Bernhardt, Chiara Marciano, Mario Rosario Guarracino

SAML-25 Workshop on Statistical and Machine Learning

Graph Neural Networks (GNNs) have established themselves as powerful tools for graph-structured data. However, when feature separability among nodes is low, conventional neighborhood aggregation strategies often result in performance degradation due to over-smoothing and noisy information propagation. In this work, we introduce a novel GNN framework that refines the aggregation process by integrating feature similarity and neighborhood entropy into node message passing. Unlike standard models that uniformly aggregate neighbor information, this new model dynamically adjusts neighbor influence, prioritizing nodes with high similarity and low entropy. We evaluate the model on synthetic graphs generated using the Stochastic Block Model (SBM), varying …

Dealing With Large Data Sets: The Data Nugget Subset Selection Approach, Vipin Kumar, Simona Balzano, Giovanni C. Porzio

SAML-25 Workshop on Statistical and Machine Learning

Analysing big data has always been a major issue because its massive volume poses significant challenges for traditional analytical techniques. When the number of instances is extremely large, existing approaches become computationally infeasible due to the complexity of many algorithms, along with memory and time constraints inherent in processing large datasets. In such cases, using a subset of the data is considered a more practical solution, and analyses are typically performed over a simple random sample drawn from the entire dataset. Various subsampling methods have been proposed to address these issues. However, they often fall short in producing representative subsamples …

A Statistical Approach To Portfolio Optimization Using Copula-Garch Models For European Investments, Jegors Fjodorovs

SAML-25 Workshop on Statistical and Machine Learning

This study explores portfolio optimization using copula functions and GARCH models, focusing on the European stock market. Traditional mean-variance methods often miss dynamic dependencies and tail risks. By applying copula-GARCH models—particularly the Student’s t copula with eGARCH—we better capture volatility asymmetries and tail dependencies. Conditional Value at Risk (CVaR) is used to evaluate downside risk across 10,000 simulated portfolios using high-performance computing. Results show that copula- GARCH models, especially eGARCH, consistently outperform traditional methods in risk-adjusted returns, offering improved risk management.

Electrochemically-Assisted Low Power Density Laser Writing On Stainless Steel Via Enrichment Of Chromium Oxides, Zechariah J. Pfaffenberger, Lianlian Liu, Tae Kyong John Kim, Vignesh Venkataramani, Lydia Kisley

Faculty Scholarship

Laser color marking produces nearly permanent, environmentally friendly, vibrant colors on surfaces. However, previous work has used high-power-density pulsed lasers to induce the physicochemical reactions for marking. Here, laser color marking on stainless steel 304 (SS304) is performed with a less expensive continuous wave (CW) laser and a power density five orders of magnitude below that previously reported by combining an electrochemical cell with a fluorescence microscope. Using a combination of optical microscopy, x-ray photoelectron spectroscopy, and bulk electrochemistry, it is demonstrated that the laser-induced luminescence and colors are due to enrichment (32 ± 9% increase) of Cr₂O₃ in the …

Measurements Lab Quiz – Exported As Brightspace Package, Vasiliy Znamenskiy

Open Educational Resources

This Brightspace-ready quiz assesses foundational measurement skills in physics laboratory settings. It includes 16 questions covering key concepts such as random and systematic errors, least counts of measurement tools (micrometer, Vernier caliper, meter stick, graduated cylinder, and triple-beam balance), and data interpretation from physical instruments. The quiz features multiple-choice and numerical answer formats, including visual instrument reading exercises via linked images. Designed for use in college-level introductory physics courses, this package supports automated grading and student feedback within the Brightspace LMS environment.

Land Instability Compounds The Risk Of Sea Level Rise In Alexandria, Egypt, Rejoice Thomas, Sara Zouriq, Shahryar Fazli, Amr Fawzy, Nikolay Grisel Todorov, Surendra Maharjan, Wenzhao Li, Erik Linstead, Daniele Struppa, Hesham El-Askary

Mathematics, Physics, and Computer Science Faculty Articles and Research

The coastal region of Alexandria Governorate in Egypt holds significant strategic importance for trade while being susceptible to extreme weather events. It confronts a dual challenge of the rising sea levels and, as found in this study, land instability. While much attention has been rightly directed towards sea level rise (SLR), the stability of the land warrants equal consideration. Here, a comprehensive analysis of land stability is conducted in Alexandria by measuring Line of Sight (LOS) displacements and assessing their topographical, hydrological, and coastal impacts. Persistent Scatterer Interferometry technique is used to measure the LOS displacements in association with land …

2025 June 5 - Tennessee Weekly Drought Summary, Tennessee Climate Office, East Tennessee State University

Tennessee Climate Office Weekly Drought Summaries

No abstract provided.

Tilings In The 3 Dimensional Lattice With L-Tetrominoes, Ian N. Bridges

Rose-Hulman Undergraduate Mathematics Journal

We consider three dimensional L-tetrominoes. We show that there exists at least one way to tile every three dimensional rectangle whose side lengths are at least $3$ and area is congruent to $1 \pmod 4$ such that one square goes untiled. In addition, we show that every three dimensional rectangle is tileable provided one side has length at least $2$ and the other is a multiple of $4$.

K-Mshc: Unmasking Minimally Sufficient Head Circuits In Large Language Models With Experiments On Syntactic Classification Tasks, Pratim Chowdhary, Peter Chin, Deepernab Chakrabarty

Computer Science Senior Theses

Understanding which neural components drive specific capabilities in mid-sized language models ($\leq$10B parameters) remains a key challenge. We introduce the $(\bm{K}, \epsilon)$-Minimum Sufficient Head Circuit ($K$-MSHC), a methodology to identify minimal sets of attention heads crucial for classification tasks as well as Search-K-MSHC, an efficient algorithm for discovering these circuits. Applying our Search-K-MSHC algorithm to Gemma-9B, we analyze three syntactic task families: grammar acceptability, arithmetic verification, and arithmetic word problems. Our findings reveal distinct task-specific head circuits, with grammar tasks predominantly utilizing early layers, word problems showing pronounced activity in both shallow and deep regions, and arithmetic verification demonstrating a …

Evaluating Vision Language Model Capabilities For Time Series Interpretation: An Empirical Study With Conversation Duration And Psychological Flourishing Data From The Studentlife Dataset, Jusung Park

Computer Science Senior Theses

This research investigates the capability of Vision Language Models (VLMs), specifically ChatGPT‑4o, to interpret and predict psychological outcomes based on visual representations of time series data. Leveraging conversation duration metrics and psychological flourishing scores from the StudentLife dataset, this study rigorously evaluates the predictive accuracy of VLMs using various methods, including zero‑shot on raw data, zero‑shot on graph data, few‑shot learning, qualitative labeling, and chain‑of‑thought reasoning. Despite multiple methodological enhancements, predictive performance remains modest, revealing significant challenges in quantitative interpretation of visualized temporal data by current multimodal models. We demonstrate that standardizing input tokens by using graph images rather than …

A Steiner Tree Vc Set System In Minor-Free (Di)Graphs, Eli Friedman

Computer Science Senior Theses

We propose a set system of maximum-covering minimum-density partial Steiner trees for planar and minor-free graphs. We show that this system has VC dimension at most h-1 for edge-weighted K_h-minor-free graphs, both directed and undirected. We also consider its geometric interpretation as a range space, proving it to be piercing.

In addition, we demonstrate how one can form a junction tree set system of bounded VC dimension from such Steiner trees. This is motivated by refining the junction tree set cover approach used in Chekuri and Jain's polylogarithmic approximation algorithm for Directed Steiner Forest in planar graphs [CJ25].

Character Relationship Prediction In Movies: Toward Emotionally-Aware Automatic Audio Descriptions, Seung Hyun Hahm

Computer Science Senior Theses

Automatic audio description (AD) systems support visually impaired audiences by narrating visual content, but they often fail to capture the interpersonal dynamics that underpin narrative understanding. In this work, we introduce a novel framework for character relationship prediction as a means of enriching audio descriptions with socially grounded context. Our contributions are threefold: (1) we propose the Character Relationship Module (CRM), which extends identity-aware video captioning with directed sentiment inference between character pairs; (2) we develop a scalable weak supervision pipeline that uses large language models to generate 669,520 relationship annotations across 202 films; and (3) we construct a complementary …

Bayesian Segmentation–Driven Informative Path Planning For Uav-Based Water Orthomosaic Generation, Phuc Dai Tran

Computer Science Senior Theses

This paper presents a comprehensive implementation

study of an informative path planning (IPP) algorithm

for autonomous water body detection and mapping using

unmanned aerial vehicles (UAVs). We propose a hybrid IPP

framework that seamlessly integrates Bayesian probabilistic

classification and real-time uncertainty quantification to achieve

superior flight efficiency and mapping accuracy compared

to conventional systematic coverage methods. Our approach

employs the state-of-the-art SegFormer deep learning segmentation

model in conjunction with log-odds-based orthomosaic

generation to produce high-fidelity water body maps under

diverse environmental conditions. Through random sampling of

the FloodNet dataset, we demonstrate that our IPP algorithm

maintains flight distance while achieving …