Digital Commons Network^™

Speculative Automated Refactoring Of Imperative Deep Learning Programs To Graph Execution, Raffi Khatchadourian Ph.D., Tatiana Castro Vélez, Mehdi Bagherzadeh, Nan Jia, Anita Raja

Publications and Research

Efficiency is essential to support ever-growing datasets, especially for Deep Learning (DL) systems. DL frameworks have traditionally embraced deferred execution-style DL code---supporting symbolic, graph-based Deep Neural Network (DNN) computation. While scalable, such development is error-prone, non-intuitive, and difficult to debug. Consequently, more natural, imperative DL frameworks encouraging eager execution have emerged but at the expense of run-time performance. Though hybrid approaches aim for the "best of both worlds," using them effectively requires subtle considerations. Our key insight is that, while DL programs typically execute sequentially, hybridizing imperative DL code resembles parallelizing sequential code in traditional systems. Inspired by this, we …

Constraints On Inner Core Composition From Seismicanisotropy: The Importance Of Light Elements And Nickel, Jada Bollmeyer, Daniel A. Frost, Prajna Paramita Das

Faculty Publications

The inner core is seismically anisotropic, with PKIKP waves traversing the inner core parallel to the rotation axis faster than those in the equatorial plane. This anisotropy increases with depth into the inner core and may result from alignment of iron crystals deformed during inner core growth. Using previously calculated elastic properties of iron, we seek to determine the most likely iron-light element alloy (FeC, FeO, FeS, or FeSi). For each FeX alloy, we interpolate elastic tensors across the pressure and temperature range of the inner core and model the anisotropy resulting from flow during core growth. Lastly, we compare …

Speculative Automated Refactoring Of Imperative Deep Learning Programs To Graph Execution, Raffi Khatchadourian Ph.D., Tatiana Castro Vélez, Mehdi Bagherzadeh, Nan Jia, Anita Raja

Publications and Research

Efficiency is essential to support ever-growing datasets, especially for Deep Learning (DL) systems. DL frameworks have traditionally embraced deferred execution-style DL code---supporting symbolic, graph-based Deep Neural Network (DNN) computation. While scalable, such development is error-prone, non-intuitive, and difficult to debug. Consequently, more natural, imperative DL frameworks encouraging eager execution have emerged but at the expense of run-time performance. Though hybrid approaches aim for the "best of both worlds," using them effectively requires subtle considerations. Our key insight is that, while DL programs typically execute sequentially, hybridizing imperative DL code resembles parallelizing sequential code in traditional systems. Inspired by this, we …

Connected-Component Labeling Using Hls For High-Energy Particle Physics Instruments, Nick Song, Marion Sudvarg, Roger Chamberlain

Computer Science Faculty Research & Creative Works

Many instruments used in high-energy particle physics observations, e.g., gamma-ray telescopes, use FPGAs for front-end signal processing of raw sensor data. The use of high-level synthesis (HLS) to express the signal processing algorithms has the potential to significantly reduce development time for new instruments of this type. We describe our experience with one of the computational stages in the signal processing pipeline, island detection, exploring its implementation across multiple configurations: 1D versus 2D islands, and 4-way versus 8-way connected-component labeling (CCL) in the 2D configuration. We report resource usage and performance for both configurations of 2D island detection, including the …

Modular Architecture For High-Performance And Low Overhead Data Transfers, Rasman Mubtasim Swargo, Engin Arslan, Md Arifuzzaman

Computer Science Faculty Research & Creative Works

High-performance applications necessitate rapid and dependable transfer of massive datasets across geographically dispersed locations. Traditional file transfer tools often suffer from resource underutilization and instability due to fixed configurations or monolithic optimization methods. We propose AutoMDT, a novel Modular Data Transfer Architecture, to address these issues by employing a deep reinforcement learning agent to simultaneously optimize concurrency levels for read, network, and write operations. This solution incorporates a lightweight network-system simulator, enabling offline training of a Proximal Policy Optimization (PPO) agent in approximately 45 minutes on average, thereby overcoming the impracticality of lengthy online training in production networks. AutoMDT's modular …

Modeling And Optimizing Real-Time Telescope Interaction For Multi-Wavelength Observation Of Gamma-Ray Bursts, Ye Htet, Marion Sudvarg, Honghao Yang, Jeremy Buhler, Roger Chamberlain, James Buckley

Computer Science Faculty Research & Creative Works

Multi-wavelength observation of gamma-ray bursts (GRBs) requires real-time interaction among multiple telescopes. A gamma-ray telescope detects and localizes a GRB in the sky and must then communicate with an optical telescope to direct the latter toward the GRB as quickly as possible. We previously developed software for ADAPT, a suborbital gamma-ray telescope, to localize GRBs in real time, on a timescale shorter than that of the GRB itself. This work therefore studies progressive localization, in which ADAPT computes a series of increasingly accurate location estimates during a GRB to enable a partner instrument to more rapidly find it. We describe …

Playa Sediment Bioaccessibility Presents Previously Unknown Risks To Population Health In A Water-Stressed Region Undergoing Lake Desiccation: Salton Sea, Usa, Jordan Jaeger

Theses

Globally, drylands are expanding, and endorheic basins are in rapid decline. Saline lake desiccation exposes lakebed playa likely to become dust, posing potential health risks to surrounding communities. Studies investigating health risks from playa dust exposure often focus on health effects associated with the presence of airborne particulates, but fewer studies investigate health risks associated with dust source chemistry. This study examines dust source particle composition in a desiccating inland sea in a water-stressed agricultural region of California. Specifically, chemical bioaccessibility of bound trace elements uncovers risks to children, adults, and agricultural worker health. Arsenic exposure via sediment ingestion poses …

Source Apportionment And Potential Health Risks Of Trace Metals In A Contaminated Urban River In New York/New Jersey Harbor System, Md Shahnul Islam, Sana Mirza, Huan Feng, Tapos Kumar Chakraborty, Yu Qian, Shinjae Yoo

Department of Earth and Environmental Studies Faculty Scholarship and Creative Works

The Lower Passaic River (LPR), located within the New York/New Jersey Harbor Estuarine System, has experienced long-term industrial activities, resulting in elevated concentrations of trace metals in sediment and water. This study aims to assess the bioaccumulation behavior, potential human health risks, and sources of copper (Cu), lead (Pb), and mercury (Hg) in the LPR. Trace metal concentrations were measured in water, sediment, and seven edible aquatic species. Data were analyzed using statistical approaches, and evaluated by bioaccumulation factors (BAFs) and human health risk assessments based on U.S. Environmental Protection Agency (USEPA) guidelines. Results showed that Hg exhibited the highest …

Chronosort: Revealing Hidden Dynamics In Alphafold3 Structure Predictions, Matthew J. Argyle, William P. Heaps, Corbyn Kubalek, Spencer Gardiner, Bradley C. Bundy, Dennis Della Corte

Faculty Publications

Protein function emerges from dynamic conformational changes, yet structure prediction methods provide only static snapshots. While AlphaFold3 (AF3) predicts protein structures, the potential for extracting dynamic information from its ensemble predictions has remained underexplored. Here, we demonstrate that AF3 structural ensembles contain substantial dynamic information that correlates remarkably well with molecular dynamics simulations (MD). We developed ChronoSort, a novel algorithm that organizes static structure predictions into temporally coherent trajectories by minimizing structural differences between neighboring frames. Through systematic analysis of four diverse protein targets, we show that root-mean-square fluctuations derived from AF3 ensembles can correlate strongly with those from MD …

Contextual Bandits For Non-Stationary Multi-Agent Environments Using Assumptions-Alignment Tracking, Ethan Pedersen

Theses and Dissertations

A common challenge in AI is that of an agent possessing N distinct behaviors, each of which works effectively in certain tasks and circumstances, to choose from at any particular time. A popular approach to learn optimal selections in such contexts is found in bandit algorithms. However, traditional bandits often (1) assume the environment is stationary, (2) focus on asymptotic performance, and (3) do not incorporate available external information about the environment. While contextual bandits that do consider external information are designed to combat one of these challenges, the others remain problematic in realistic domains. This is especially true of …

Reducing Data Requirements In Polymer Science: Deep Neural Networks For Predicting Surface Tension Of Copolymer Compatibilizers, Md Mushfiqul Islam

USF Tampa Graduate Theses and Dissertations

In polymer chemistry, compatibilization involves adding a substance often a block or graft copolymerto stabilize polymer blends that would otherwise not mix well, leading to rough structures and weak me- chanical properties. Compatibilizers improve miscibility and reduce interfacial tension, which is critical for applications such as mixed-waste polymer recycling. Sequence-controlled polymers offer unique potential by combining the tunable chemistry of synthetic polymers with the precise, function-driven design of biological macromolecules, but unlike proteins, they lack large, evolution-shaped datasets to guide discovery. This research develops a deep learning framework to predict the surface tension of sequence-controlled copolymer compatibilizers across varying concentrations. …

Speculative Automated Refactoring Of Imperative Deep Learning Programs To Graph Execution, Raffi T. Khatchadourian Ph.D., Tatiana Castro Vélez, Mehdi Bagherzadeh, Nan Jia, Anita Raja

Publications and Research

Efficiency is essential to support ever-growing datasets, especially for Deep Learning (DL) systems. DL frameworks have traditionally embraced deferred execution-style DL code---supporting symbolic, graph-based Deep Neural Network (DNN) computation. While scalable, such development is error-prone, non-intuitive, and difficult to debug. Consequently, more natural, imperative DL frameworks encouraging eager execution have emerged but at the expense of run-time performance. Though hybrid approaches aim for the "best of both worlds," using them effectively requires subtle considerations. Our key insight is that, while DL programs typically execute sequentially, hybridizing imperative DL code resembles parallelizing sequential code in traditional systems. Inspired by this, we …

Classifications Of Gradient Almost Ricci Solitons Within The Framework Of Closed Conformal Vector Fields, Akram Ali, Fatemah Mofarreh, Jae Won Lee

Turkish Journal of Mathematics

In this paper, we consider gradient almost Ricci solitons with nonparallel closed conformal vector fields. We classify gradient almost Ricci solitons and show that a complete, noncompact gradient almost Ricci soliton admitting a nonparallel closed conformal vector field, is locally a warped product with Einstein fibers, while in the second case, under additional hypothesis, gradient almost Ricci soliton is totally umbilical if and only if it has constant scalar curvature.

Artificial Intelligence (Ai) And The Anthropic Economic Index In The Mountain West, 2025, Cason Noll, Olivia K. Cheche, William E. Brown Jr.

Economic Development & Workforce

This fact sheet presents 2025 data on the state of artificial intelligence (AI) adoption among the five Mountain West states of Arizona, Colorado, Nevada, New Mexico, and Utah. The data are sourced from the “Anthropic Economic Index,” which provides data on Claude.ai (an AI large language model) and its adoption across all 50 U.S. states and Washington, D.C. This fact sheet focuses on Claude.ai usage, the most common topic Claude.ai has been used for, and augmentation and automation shares for each Mountain West state.

Exploratory Study Of Semiconductor Nanomembranes In Em Applications, Grant D. Heileman

Electrical and Computer Engineering ETDs

Antenna systems are a cornerstone of modern technologies, playing an increasingly vital role in their advancement. As demand for compact, high-performance, and adaptable communication platforms grows reconfigurable antenna technologies are becoming essential. This research explores a novel front-end reconfigurable antenna system (FERAS) architecture that leverages the mechanical flexibility and photoconductive behavior of semiconductor nanomembrane (SNM) devices. By exploiting the emergent properties of ultra-thin silicon (Si) or gallium arsenide (GaAs) nanomaterials and optically exciting these samples using vertical-cavity surface-emitting laser (VCSEL) arrays, this study develops lightweight, low-cost, deployable antenna structures for satellite communications, remote sensing, GPS, and radar. Despite their significant …

Software-Defined Networking Powered By Ai-Driven Anomaly Detection, Dina Moloja, Vusumuzi Malele Prof

Journal of Cybersecurity Education, Research and Practice

Software Defined Networking (SDN) revolutionizes network control by separating the control plane from the data plane. Although the latter improves SDN agility and scalability, it creates a security hole, particularly in a central control plane, leading to SDN environments becoming high-profile targets for advanced cybersecurity threats. Due to static and signature-based point-in-time behavior, traditional security methods are unable to keep up with modern attacks that are an anomaly to SDNs. Artificial Intelligence (AI) with its different applications and techniques, has the capability of detecting SDN cyber threats’ anomalies. This paper presents the results of a literature scoping exercise that used …

Locations Of Outfalls In The Indian River Lagoon And Lake Worth Lagoon, Florida, Kelly M. Kibler, Melinda J. Donnelly, Vincent Encomio, Madison Giuntoli, Manisha Thenuwara, Brice Bennett, Laura Andrade Barron, Mariam Sonbol, Wynter Raine Seppala, Namritha Ramakrishnan, Jennifer Jordan Báez, Beth Orlando, Irene Arpayoglou, Melissa Meisenburg

Indian River Lagoon Shorelines

The spatial dataset: UCF_Outfalls_IRL_LWL.shp was created under grants from the Indian River Lagoon National Estuaries Program and Florida Sea Grant. These projects cumulatively led to creation of a living shoreline restoration prioritization model and hydrodynamic habitat suitability models for 1,550 km (963 miles) of estuarine shorelines in Indian River Lagoon (IRL) and Lake Worth Lagoon (LWL).

The dataset presented here includes the location of outfalls identified throughout the study region and general characteristics of the shoreline where they were identified.

Correcting Class Imbalance Through Synthetic Training Data And 3d Modeling For Carabid Pitfall Trap Sampling, Blair Mirka

Geography ETDs

Crowdsourced biodiversity data provide an accessible foundation for large-scale ecological monitoring, but class imbalance limits automated species identification, particularly for rare taxa. This research explores the use of synthetic training data generated from 3D models of carabid beetle museum specimens to improve detection and classification performance for underrepresented species in crowdsourced datasets. High-resolution 3D models were created to simulate variation in lighting, orientation, and background. These synthetic images were incorporated into convolutional neural network training datasets at varying synthetic-to-real ratios to assess their impact on classification accuracy. Models were evaluated using controlled pitfall-trap imagery to examine the influence of scene …

Soccer In-Game Event Classification Using Spatio-Temporal Data, Million Haileyesus

Theses and Dissertations

Classifying soccer ball events, such as pass, shot, ball loss, and ball out, are crucial for advancing game analytics and tactical insights. This thesis investigates the application of machine learning to classify these ball events using player and ball spatio-temporal data, as well as additional features. We implement and compare traditional machine learning algorithms (AdaBoost, Logistic Regression, and Random Forest) with several deep learning approaches, including Feed-Forward Neural Network (FFNN), sequence-to-sequence (seq2seq) recurrent models (Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU)), and Transformer. Our experiments, evaluated on a dataset comprising of three professional soccer matches using accuracy, precision, …

Proactllm: Proactive Conversational Information Seeking With Large Language Models, Shubham Chatterjee, Xi Wang, Shuo Zhang, Sajad Ebrahimi, Zhaochun Ren, Debasis Ganguly, Gareth Jones, Emine Arrousse, Hamed Zamani

Computer Science Faculty Research & Creative Works

Large Language Models (LLMs) have transformed information access by enabling human-like text understanding and generation. This workshop explores the next step for conversational AI: building proactive information-seeking assistants that go beyond reactive question answering. We aim to investigate how LLMs can anticipate user needs, model complex context, support mixed-initiative interactions, integrate retrieval and external tools, personalize responses, adapt through feedback, and ensure fairness, transparency, and cognitive grounding. Bringing together experts from NLP, IR, HCI, and cognitive science, the workshop will serve as a timely forum for advancing intelligent, proactive dialogue systems. It will also foster interdisciplinary collaboration.

Multiplexed Integrin Detection And Cancer Cell Classification Using Multicolor Gap-Enhanced Gold Nanorods And Machine Learning Algorithm, Suprava Shah, Reed Youngerman, Alberto Luis Rodriguez-Nieves, Mitchell Lee Taylor, William Rodney Bantom, David Thompson, Jingyi Chen, Yongmei Wang, Xiaohua Huang

Chemistry & Biochemistry Faculty Publications and Presentations

Integrins, cell-surface adhesion receptors involved in tumor progression, invasion, and metastasis, serve as crucial biomarkers for cancer diagnosis and therapeutic targeting. Multiplexed detection of integrins and cancer cell classification at the single-cell level allows for comprehensive profiling, facilitating precise identification and categorization of tumor cells that are heterogeneous in integrin expression and cell subtype. In this study, we developed a five-plex detection platform and demonstrated integrin profile for cancer cell classification leveraging surface-enhanced Raman scattering (SERS) with gap-enhanced gold nanorods (GENRs) in conjunction with advanced computational analysis. Specifically, we synthesized GENRs bearing five distinct Raman nanotags, each producing a unique …

New Evidence Of Mid-Pleistocene Glaciations In The French Jura Mountains Using Debris Flow Deposits And U/Th Dating Of Speleothems (Verneau Karst Network), Margot Vivier, Stéphane Jaillet, Eglantine Husson, Edwige Pons Branchu, Jean-Baptiste Charlier

International Journal of Speleology

We redefine the extent of former glaciers in the northern Jura Mountains through the study of endokarst deposits. Sedimentary deposits (pebbly debris flows, laminated clays, and allochthonous limestone blocks) and chemical deposits (speleothems) were identified and studied using three complementary approaches: (i) detailed geomorphological mapping, (ii) petrographic analysis, and (iii) U/Th-dating. In addition, a surface geomorphological analysis made it possible to calculate the minimum erosion time required for the formation of the sinkhole linked to the underground study site. The study site is the Dolois Chamber, an underground site upstream of the Verneau karst network (Doubs Department, France). The combined …

An Analysis Of Large Language Model Hallucination And Creativity: Linear Representations And Causal Roles, Mckay Bowcut

Theses and Dissertations

Large language models (LLMs) show remarkable abilities in both factual and creative tasks, yet the problem of hallucination persists despite advances in training and post-training methods. Recent progress in model interpretability suggests that model behavior can be predicted and influenced by analyzing and manipulating internal activations. In this thesis, we address three questions: (1) Can we identify linear representations of hallucination and creativity in the model’s latent space? (2) Do these representations play a causal role in model behavior when manipulated? (3) Are creativity and hallucination causally intertwined? To answer these, we construct a novel dataset of factual, hallucinated, and …

Sound Practices In Human Resources: A Critical Analysis Of How Ai Applicant Screening Tools Are Used With Deaf Individuals In The Hospitality And Tourism Sector, Sherry L. Gaynor

USF Tampa Graduate Theses and Dissertations

This study explores the intersection of Career and Workforce Education, Deaf Education, and artificial intelligence (AI) technologies used in hiring practices. The purpose of this research is to identify potential barriers to equitable employment for Deaf and Hard-of-Hearing (DHH) individuals and to propose effective methods for evaluating and assessing AI and machine learning applications utilized in recruitment and selection processes. Although AI is increasingly embedded in human resource management (HRM), there remains a significant lack of research addressing its use with special populations. In particular, no prior studies have examined how voice recognition software, facial analysis, psychometric testing, and virtual …

Statistical Challenges And Simulation Results For Pilot Clinical Trials, Weiliang Cen

USF Tampa Graduate Theses and Dissertations

Background: The effect size estimated from a pilot trial is often an inaccurate reflection of the true effect size observed in a large trial, leading to either underestimation or overestimation. Published data suggest that effect sizes from large trials are typically smaller than those reported in their corresponding pilot trials. To address this discrepancy, conservative or discount adjustment methods are widely recommended to modify pilot trial effect sizes when calculating sample sizes, thereby maintaining adequate statistical power. This study aims to assess effect sizes from both pilot and large trials and to evaluate the performance of existing adjustment methods.

Methods: …

From The Editors, Rully Karim Dr.

Journal of Project Management & Construction

The Journal of Project Management and Construction (JPMC) is a peer-reviewed publication dedicated to advancing the field of project management and construction, grounded in the principles outlined in the PMBOK 6th Edition. Our focus encompasses the ten knowledge areas essential to successful project management: Integration, Scope, Schedule, Cost, Quality, Resource, Communications, Risk, Procurement, and Stakeholder Management.

JPMC publishes original research papers written in English that provide deeper insights into these knowledge areas and contribute to the development of best practices in project management and construction. Submissions may include theoretical analyses, computational models, experimental observations, or a combination of both theoretical …

On Heat Engines Having Efficiencies That Degrade With Every Cycle That Are Nonetheless Reversible: The Wonders Of Hilbert Curves In The Pressure-Volume Plane, Daniel Graham

Chemistry: Faculty Publications and Other Works

Heat engines are cornerstones in classical thermodynamics, not to mention fixtures of everyday life. As everyone knows, real heat engines are rife with sources of irreversibility, and, in turn, have efficiencies that degrade over time. Here we chart a classical heat engine whose efficiency degrades with every cycle, but maintains perfect reversibility nonetheless. There are no paradoxes in play due to the inviolability of the thermodynamic laws. Rather the engine operates transparently according to Hilbert curves that fill the pressure-volume plane. The assembly of the curves is iterative and digital whereby no information is lost over every cycle.

Development Of Problem-Based Learning Worksheet Toenhance Ecoliteracy: Human Impact On Ecosystem, Ulina M.S. Simangunsong, B. Manurung, Martina Asiati Napitupulu

Jurnal Pendidikan Sains

Ecoliteracy (ecological literacy) is an individual's ability to understand basic ecological principles and make environmentally responsible decisions. Improving ecoliteracy is a crucial foundation for realizing Education for Sustainable Development (ESD), which prepares students to become proactive citizens in protecting the environment. The problem-based learning model helps students learn on their own by using real-world environmental cases to get them to think critically and solve problems. Therefore, this research aims to determine the effectiveness of using problem-based learning worksheets to enhance the ecoliteracy of seventh-grade students. The development model used is 4-D. The research instruments include expert validation forms for subject, …

Site Specific Reliability-Targeted Snow Loads And Winter Wind Parameters Across The World, Brennan Bean, Nicholas Brimhall, Bikram Bhusal, Marc Maguire, Maha Moussa

Mathematics and Statistics Faculty Publications

This report presents a framework for estimating Reliability Targeted Snow Loads (RTSLs) and Winter Wind Parameters (WWPs) at locations outside the Conterminous United States (OCONUS). The methodology integrates ground-based in-situmeasurements with gridded global climate products to generate spatially continuous RTSL estimates for nearly any location in OCONUS. RTSLs are computed at qualifying in-situ stations, and predictive models relating gridded climate products to in-situRTSLs enable estimation at locations lacking direct measurements. This report describes the development of the in-situ annual maximum snow load database, the construction of global gridded climate summaries, the estimation of site-specific RTSLs and Service Targeted …

Studies Of Seismicity In The Alaska Subduction Zone And Antarctica, Patrick Matulka

Arts & Sciences Graduate Student Theses and Dissertations

This dissertation utilizes multiple methods of earthquake detection, relocation, and characterization to investigate the patterns of seismicity along the Alaska subduction zone and Antarctica. The studies of the Alaska subduction zone are built on the data collected during the Alaska Amphibious Community Seismic Experiment (AACSE) that deployed temporary land and ocean-bottom seismometers along the Alaska Peninsula and offshore regions from May 2018 to September 2019. Using seismic arrivals at those stations, as well as a combination of other land-based networks, I relocated 1307 earthquakes in the incoming plate and trench slope regions of the Alaska subduction zone with an absolute …