Digital Commons Network^™

A Uas-Centered Investigation Of Vorticity Characteristics And Cold Pool Structure Across Forward And Left-Flank Boundaries In Supercells, Mark R. De Bruin

Department of Earth and Atmospheric Sciences: Dissertations, Theses, and Student Research

Supercell internal boundaries are the locus of tornadogenesis; thus, understanding the characteristics of these boundaries, particularly in terms of vorticity, is important for identifying the role they play in tornado formation. Insight into the overall characteristics of internal boundaries and their possible role in tornadogenesis have been driven by studies reliant on numerical modeling-based experiments. Observational studies often neglect above-surface conditions or, when these observations are made, lack the spatial resolution to resolve boundary characteristics. During TORUS (Targeted Observation by Radars and UAS of Supercells) 2019 and TORUS-LItE (TORUS Left-flank Intensive Experiment) 2023, uncrewed aircraft systems (UAS) and mobile mesonets …

Statistical Monitoring Of Hard Faults In Digital Systems, Dany Akshay Deep Isukapalli

Electrical Engineering Theses and Dissertations

Achieving a high test coverage is crucial for helping to ensure that integrated circuits are working correctly and are non-defective. Although scan-based structural tests are used throughout the industry, high-level functional tests may be needed to detect some defects— especially those that are environmentally sensitive. Unfortunately, the character of functional test makes it difficult to obtain high coverage, and it is even hard to estimate coverage because fault simulation times of large circuits are long. As a result, some method is required for predicting the ability of a functional test that has not been fault simulated to detect defects. In …

Towards Efficient Privacy-Preserving Deep Learning: He-Friendly Structures, Flexible Pruning, He-Efficient Architectures, And Secure Transformer Token Drop, Yifei Cai

Electrical & Computer Engineering Theses & Dissertations

Deep learning (DL) has become a powerful tool for solving complex problems, but developing DL models typically requires vast datasets, high computational resources, and expert knowledge—barriers that limit accessibility. Machine Learning as a Service (MLaaS) addresses this challenge by allowing resource-rich providers to deliver pre-trained DL models as services. However, privacy concerns arise: clients hesitate to share sensitive data, while providers protect their proprietary models. To address this, privacy-preserving MLaaS integrates cryptographic techniques into DL computations, as seen in frameworks like Cryptonets, SecureML, GAZELLE, CrypTFlow2, Cheetah, and BOLT. Among them, Homomorphic Encryption (HE) enables computation on encrypted data but remains …

Advancing Food Nutrition Estimation Via Visual-Ingredient Feature Fusion, Huiyan Qi, Bin Zhu, Chong-Wah Ngo, Jingjing Chen, Ee-Peng Lim

Research Collection School Of Computing and Information Systems

Nutrition estimation is an important component of promoting healthy eating and mitigating diet-related health risks. Despite advances in tasks such as food classification and ingredient recognition, progress in nutrition estimation is limited due to the lack of datasets with nutritional annotations. To address this issue, we introduce FastFood, a dataset with 84,446 images across 908 fast food categories, featuring ingredient and nutritional annotations. In addition, we propose a new model-agnostic Visual-Ingredient Feature Fusion (VIF2 ) method to enhance nutrition estimation by integrating visual and ingredient features. Ingredient robustness is improved through synonym replacement and resampling strategies during training. The ingredient-aware …

Secure Query On Encrypted Data By Using Fully Homomorphic Encryption, Mrinmoy Bera

Master’s Dissertations

Cloud service providers typically store user data in an encrypted form (data at rest). However, when a user performs a query, the server first decrypts the data, processes the query on plaintext, and then sends the result back to the user (data in transit). This process exposes a critical vulnerability—if the cloud server is ever compromised, the decrypted data becomes accessible to the attacker. To address this security gap, we design a secure query protocol that eliminates the need to decrypt data on the server side. Fully Homomorphic Encryption (FHE) offers a groundbreaking solution by enabling arbitrary computations directly on …

Federated Learning Using Fully Homomorphic Encryption, Sk Golam Kuddus

Master’s Dissertations

Traditional machine learning approaches require centralizing data for training, which raises significant privacy concerns when dealing with sensitive information. Federated learning (FL) addresses this by keeping data local and enabling multiple users to collaboratively train a shared machine learning model. In spite of this, FL remains vulnerable to inference attacks, as sensitive information can still be extracted from the model’s learned parameters. While traditional privacy-enhancing techniques such as di!erential privacy introduce noise to model updates to obscure individual data points, they often present a fundamental trade-o! between privacy and utility. Furthermore, these approaches still carry risks of data leakage if …

Modeling Multiple Tasks In Recommendation Systems, Dinh Hieu Do

Dissertations and Theses Collection (Open Access)

Traditional research in recommendation systems has largely centered on the static offline supervised learning setting. In this paradigm, all available user-item interaction data is collected and partitioned into fixed training, validation, and test sets. Models are developed and evaluated in this controlled environment, where the underlying data distribution is assumed to remain unchanged. This approach offers clear advantages: it simplifies experimentation, enables reproducible benchmarking, and allows for straightforward comparisons between algorithms.

However, this static offline setting does not reflect the realities faced by modern recommendation systems. In real-world applications, data is dynamic and ever-evolving, where new users and items are …

From Sparse Feedback To Sequential Decision-Making: Learning Safety Constraints With Weak Supervision, Siow Meng Low

Dissertations and Theses Collection (Open Access)

Real-world decision-making often involves safety constraints that are implicit, non-Markovian, or difficult to specify directly. Standard reinforcement learning (RL) approaches typically assume access to fully specified cost functions and constraint budgets—assumptions that limit their applicability in domains where such structure must instead be inferred from data. This dissertation develops a sequence of methods for learning safety-relevant structure from weak supervision, such as sparse binary feedback on trajectory segments, and using these signals to guide planning and policy optimization.

The first part of the dissertation introduces a sample-efficient method for planning in continuous Markov Decision Processes (MDPs) using deep reactive policies. …

Analysis Of Vision Transformers And Domain Adaptation In Long-Range Facial Recognition, Zachary Michael Swanson

Department of Electrical and Computer Engineering: Dissertations, Theses, and Student Research

Atmospheric turbulence presents a significant barrier to long-range facial recognition, introducing severe geometric distortions and blur that degrade image quality. This thesis investigates deep learning approaches for mitigating these effects, with a focus on transformer based architectures and domain adaptation strategies.

An in-depth benchmarking study was performed using convolutional neural networks (CNNs) and vision transformers (ViTs) on the Husker BRIAR Research Collection from up to 500m (HBRC-500) face dataset. The results demonstrated that vision transformers, particularly hierarchical vision transformers like the shifted-window (Swin) transformer, outperform CNN-based models at long distances due to their ability to model global spatial relationships and …

Multi-Modal Covid-19 Detection Using Cough Sounds And Medical Information With Attention-Enhanced Deep Learning, Mohamed Talaat Saidahmed, Reda Elbasiony, Marwa Reda Bastwesy, Asmaa Aly Hagar

Journal of Engineering Research

The COVID-19 pandemic has highlighted the need for fast, non-invasive, and cost-effective diagnostic tools. Cough sounds, as a prominent symptom of respiratory diseases, present a promising modality for automated COVID-19 detection. In this study, we propose a novel multi-modal deep learning framework for COVID-19 detection that leverages cough sounds and patient-specific medical information. Our approach extracts two types of acoustic features—Mel-Frequency Cepstral Coefficients (MFCCs) and Mel spectrograms—and integrates them with clinical metadata to improve diagnostic ac-curacy. The MFCC branch employs 1D convolutional layers followed by Efficient Channel Attention mechanism. The Mel spectrogram branch utilizes ResNet-50 combined with ECA to capture …

A Hybrid Binary Grey Wolf Optimizer Based On De Algorithm For Feature Selection, Amany Saad Abdelrazek, Basma Ghareeb Elkilany, M. Arafa

Journal of Engineering Research

Feature selection is one kind of optimization problem that has bio-objective functions, where it is necessary to get the minimum number of features that achieve high classification accuracy. According to literature studies, several kinds of meta-heuristic algorithms have been utilized to solve feature selection problems. One of these meta-heuristic algorithms is the Grey Wolf Optimization (GWO) algorithm and its modified variants, including the binary versions. They have yielded competitive results compared to other algorithms. Despite the simplicity and effectiveness of GWO and its modified versions, they face challenges related to the exploitation ability of the local search. To avoid premature …

Evaluation Of Machine And Deep Learning Models For Predicting Water Distillate Rate, Ghada Hamisa

Journal of Engineering Research

Freshwater scarcity has become a critical global challenge due to rapid population growth and environmental pollution caused by industrial and urban expansion. Solar stills offer a sustainable solution by desalinating impure water using solar energy, making them valuable for domestic, industrial, and academic applications. However, traditional methods for optimizing solar still performance face significant limitations, including time-consuming experimental data collection, computational inaccuracies, and high development costs. To address these challenges, this study leverages machine learning (ML) and deep learning (DL) techniques to predict the distilled water production rate of solar stills before physical construction or modification. A heat pump solar …

Efficient Methods And Algorithms For Analyzing Stochastic Systems, Mohammad Ahmadi

USF Tampa Graduate Theses and Dissertations

This dissertation addresses the challenges of stochastic analysis of safety-critical systems with biological components, where unexpected behavior can lead to catastrophic events. Two fundamental challenges hinder the analysis of such systems: their typically large or infinite state spaces, and the extreme rarity of error states of interest. While Monte Carlo simulation can analyze biochemical systems without storing the state space, accurately estimating rare event probabilities becomes computationally prohibitive. Conversely, probabilistic model checking excels at analyzing extremely low probability events but becomes impractical for systems with large or infinite state spaces due to memory constraints.This work proposes two main contributions to …

Using Blockchain Technology To Help Secure America's Defense Critical Infrastructure, Vimal Buck, Aerin Krebs, Brynn Hillard, Jakob Gerha, Joseph Lutma, Srikar Maduposu, Ted Allen

Military Cyber Affairs

Critical water infrastructure in the United States faces increasing cybersecurity threats from state-sponsored actors, with potentially devastating consequences for national security, economic stability, and public health. (Cybersecurity and Infrastructure Security Agency, 2025). This infrastructure supports defense critical assets and is actively being targeted by various state-sponsored hacking groups, which poses a major concern for civilians and military alike. K. Herath (personal communication, February 24, 2025) reported being aware of two attacks on Ohio water systems during his tenure as Cybersecurity Strategic Advisor to Ohio Governor Mike DeWine.

Water is essential to everyday life and defense and presents as a high-value …

Quantifying Adversary Military Forces’ Susceptibility To Cognitive Attacks, Bonnie Rushing, Cole Nelson, Shouhuai Xu, Christofer “Raven” O’Keefe, Olga Karpoyan

Military Cyber Affairs

This paper introduces a method to quantify international populations’ susceptibility to cyber cognitive attacks using press freedom and media trust metrics. We present the Cognitive Influence Calculator, a tool that estimates susceptibility (𝑆) based on Press Freedom Scores (PFS) and media trust levels. Findings show that while authoritarian regimes are harder to reach, successful cognitive attacks have greater impacts due to higher trust in state-controlled narratives. Using U.S. wargaming data and international trust metrics, we compute susceptibility scores for the U.S., Russia, China, Iran, and North Korea. Results show an inverse relationship between PFS and media susceptibility, with local/allied …

Tweaking Ml-Kem (Kyber) And Ml-Dsa (Dilithium), Kumar Rahul

Master’s Dissertations

Lattice-based cryptography is the use of conjectured hard problems on point lattices in Rn as the foundation for secure cryptographic systems. Attractive features of lattice cryptography include apparent resistance to quantum attacks (in contrast with most number-theoretic cryptography), high asymptotic efficiency and parallelism, security under worst-case intractability assumptions, and solutions to long-standing open problems in cryptography. This work surveys the structure, security, and optimization potential of two leading lattice-based cryptographic schemes: ML-KEM (Kyber) and ML-DSA (Dilithium). Special attention is given to their applicability in government-oriented post-quantum cryptographic systems, focusing on performance, implementation considerations, and resilience against known quantum threats. In …

Context-Switch Attacks: Understanding And Mitigating The Threat To Llm Applications, Sydney Holder, Bivin Sadler

SMU Data Science Review

Large Language Models (LLMs) are transforming conversational AI, yet their dependence on prompt-supplied context exposes them to context-switch attacks that covertly steer dialogue toward sensitive or malicious ends. A 70 one-sided conversation transcript evaluation set was constructed spanning various fraudulent scenarios. Each transcript embeds adversarial patterns drawn while preserving natural conversational flow. We introduce a hybrid defense that pairs a BERT-based semantic-drift detector (cosine-similarity threshold = 0.70) with a curated keyword and hack-phrase scanner to counter these threats. In aggregate, the system delivered 100 % recall, intercepting every simulated phishing or data-harvesting attempt. The keyword layer achieved perfect precision, generating …

Finite-Time Robust Anti-Disturbance Control For Steer-By-Wire System, Jingyi Zhang, Xin Chen, Jingang Ding, Jianguo Luo, Shuo Feng

Journal of System Simulation

Abstract: To eliminate the influence of parameter perturbations and external disturbances on the wheel angle tracking control performance of steer-by-wire (SbW) system, a fractional-order integral terminal sliding mode control scheme based on a finite-time disturbance observer is proposed. A sliding modebased second order finite-time disturbance observer (FDO) is designed to precisely estimate the total disturbance of the SbW system, and the estimated total disturbance is compensated into the system control input to reduce the wheel angle tracking error. A fractional-order fast integral terminal sliding mode control (FOFITSMC) scheme is designed to ensure fast convergence of the wheel angle tracking error …

Research On Obstacle Avoidance Of Substation Robot Based On Spatiotemporal Networks, Chong Cheng, Lixia Wang, Songtao Duan, Xiaoguang Xiong, Xianjun Ge

Journal of System Simulation

Abstract: In order to improve the visual obstacle avoidance ability of substation robots in complex environments, a robot visual obstacle avoidance method based on spatiotemporal networks is proposed. The method utilizes traditional image processing techniques to enhance road information and designs a lightweight deep convolutional neural network structure to extract road features from a spatial domain perspective; based on the spatial characteristics of the road, a long short-term memory network is introduced to mine the changes in the road from a temporal perspective, and a classification regression prediction structure is used to predict the robot's obstacle avoidance direction and angle; …

Research On Scenario-Driven Virtual Simulation Test Method For Autonomous Escort Function Of Habor Tugs, Shijie Li, Jialin Li, Jialun Liu, Chengqi Xu, Zhilin Dong

Journal of System Simulation

Abstract: In order to comprehensively construct the test scenarios and verify the reliability of the tugboat autonomous companionway function, a scenario-driven virtual simulation test method for the tugboat autonomous companionway function is proposed. Based on the relative heading, relative speed and relative position of the target ship and the tugboat, the test cases of the tugboat autonomous companionway scenario are generated, and the complexity of the test cases is evaluated by using the fifthorder Bessel curve. The autonomous companion navigation function of the tug is verified through simulation experiments on the complex typical test scenarios without and with obstacles. The …

Self-Supervised Defect Detection Via Discriminative Enhancement-Based Distillation Learning, Zhiyuan Feng, Ying Chen

Journal of System Simulation

Abstract: To address the issues of scarce and unknown types of abnormal defect data and the lack of diversity in anomaly representation in conventional knowledge distillation defect detection methods, a self-supervised distillation learning method based on discriminative enhancement is proposed. An attention-based multi-scale feature fusion module is proposed, which enhances the capability of anomaly representation by amplifying the multi-scale feature differences between the student network and the teacher network. A discriminative network composed of a feature reweighting module and a decoder is designed to generate more accurate anomaly score maps by further emphasizing the anomaly features in the teacher network, …

Research On Robot Dynamic Obstacle Avoidance Method Based On Improved A* And Dynamic Window Algorithm, Yan Zhang, Binghua Li, Tao Huo, Rong Liu

Journal of System Simulation

Abstract: Aiming at the problems that the traditional A* algorithm has too many extension nodes and path turning points, and can't deal with dynamic obstacles in complex environment, a robot obstacle avoidance method combining improved A* algorithm and DWA algorithm is proposed. The A* algorithm improves the neighborhood expansion method and effectively avoids the problem of redundant nodes in the classical four-neighborhood expansion and the path through the obstacle in the eight-neighborhood expansion. A quadrant selection method is proposed, which can effectively reduce the number of extended nodes in the path search process. The redundant point elimination strategy is proposed …

Modeling And Simulation Of Hybrid Traffic Flow Considering The Inherent Dynamics Of Cacc Vehicular Platoons, Xiujian Yang, Jingjing Huang, Xi Wang

Journal of System Simulation

Abstract: To investigate the characteristics of single-lane mixed traffic flow with the presence of cooperative adaptive cruise control (CACC) vehicle platoons, a modeling approach based on cellular automata is proposed. This method distinguishes between the car-following strategies of human-driven vehicles and CACC vehicles, incorporating dynamic inter-vehicle spacing within the platoon and actual control behaviors to construct a mixed traffic flow model with inherent dynamic properties. The model enables an in-depth analysis of the influence of platoon features, such as geometric formation, carfollowing control strategies, and platoon size, on the characteristics of mixed traffic flow. It also allows us to study …

Automatic Multi-Objective Optimization Based On Dynamic Storage Location Allocation Strategy, Juan Chen, Wang Zheng, Qianqian Liu, Bin Lu

Journal of System Simulation

Abstract: Based on the dynamic storage allocation strategy, the two-stage optimization model is constructed with the whole warehouse as the main optimization body, in order to meet the safety and rationality of the storage allocation goals, and to meet the dispatching goals of the shortest operation time and the lowest energy consumption of each stacke. The upper and lower levels of the model are typical multi-objective optimization problems, and the ideal solution of the upper level model will be the initial condition of the lower level model. The multi-objective genetic algorithm is used to solve the ideal solution of the …

Soft Sensor Modeling Based On Adaptive Sparse Broad Learning System⋅, Kangping Du, Lin Sui, Weili Xiong

Journal of System Simulation

Abstract: To address the challenges posed by nonlinearity and the coupling of multiple features in complex industrial processes, resulting in increased model complexity and decreased performance, a soft sensor modeling method based on adaptive sparse broad learning system is proposed. Building upon the lateral enhancement transmission of features, the trace least absolute shrinkage and selection operator (LASSO) is further used to optimize the feature weights of the network, adaptively adjusting the penalty intensity based on the correlation between different variables to enhance the feature extraction capabilities of the model. The Dropout mechanism is introduced in the enhanced part, and the …

Enhanced Artificial Gorilla Algorithm For Mobile Robot Path Planning, Chen Ye, Peng Shao, Shaoping Zhang, Wenting Li, Tengming Zhou

Journal of System Simulation

Abstract: To address the issues of susceptibility to local optima and slow convergence in mobile robot path planning within complex terrain scenarios, an enhanced artificial gorilla troops optimizer with integration of quadratic interpolation and elite individual genetic strategies (QGGTO) is proposed. The algorithm integrates quadratic interpolation and elite individual genetic strategies to promote information exchange among candidate solutions, thereby accelerating convergence, while maintaining population diversity to avoid local optima. For complex terrains containing both regular and irregular obstacles, a cost function that comprehensively considers walking distance, safety, and turning angles is constructed to uniformly evaluate the path planning performance of …

Multi-Model Based Iterative Method For System-Of-Systems Architecture Design, Xuemeng Zhao, Tianzhu Ren, Zhemei Fang

Journal of System Simulation

Abstract: In order to solve the problems of difficulties in expressing dynamic characteristics and lack of decision analysis support in developing models of the department of defense architecture framework (DoDAF), an integrated iterative method for combat SoS architecture design is proposed. The DoDAF architecture model integrates and expresses combat-related information from multiple perspectives; the ExtendSim executable model simulates the emergence behavior and dynamic characteristics of combat SoS architecture in multiple scenarios; and the decision model quantitatively analyzes and selects architecture schemes by multi-objective decision rules. Ultimately, a SoS architecture integrated iterative design method of "view-simulate-decide-iterate" is formed. The design process …

A Drl⁃Based Approach For Distributed Equipment Nodes Selection, Ziyi Wang, Kai Zhang, Dianwei Qian, Yuzhen Liu

Journal of System Simulation

Abstract: Aiming at the problem of insufficient solution speed and poor generalization of traditional algorithms in large-scale scenarios, this paper intelligently solves the large-scale distributed equipment system preference problem based on deep reinforcement learning. According to the characteristics of distributed equipment system combat, using the complex network to its graph form modeling, and based on the attention mechanism to the equipment between the connecting edge relationship for the characterization, in order to build a distributed equipment system digital simulation environment. Simulation results show that compared with the genetic evolutionary algorithm, the obtained model has obvious advantages in terms of solution …

Simulation Study On Adaptive Signal Control Of Deformed Intersection Based On Lstm-Gnn, Kun Chen, Liang Chen, Jiming Xie, Fengbo Liu, Taixiong Chen, Lukuan Wei

Journal of System Simulation

Abstract: Aiming at the traffic congestion at deformed intersections, an improved adaptive traffic signal control scheme based on deep learning is designed, the scheme integrates the adaptive signal control of LSTM and GNN at deformed intersections. LSTM is used to capture the dependence between time series traffic data, while GNN is used to construct a spatial interaction model between lanes. By integrating the information of time and space dimensions, the model can dynamically adjust the phase duration of signal lights according to real-time traffic conditions. The results indicate that the LSTM-GNN adaptive control scheme improves overall traffic throughput efficiency by …

Modeling And Simulation Of Dual-Podded-Propulsion Ship Motions, Bing Han, Yunhe Lin, Yuhang Chen, Zhouhua Peng

Journal of System Simulation

Abstract: Aiming at the autonomous navigation control requirements of the Dalian Maritime University's dual-purpose intelligent research and training ship "Xin Hong Zhuan," the design of the motion model for this dual-podded-propulsion ship is carried out. Utilizing an MMG model structure, it calculates the hull's hydrodynamic viscous forces, single/dual-propeller thrust, and hydrodynamic forces acting on the podded propulsion units. Based on data from sea trials and open-water propeller tests, straight-navigation resistance is derived via data fitting, while a method using simulated turning circle tests and PSO algorithms is proposed to determine some hydrodynamic coefficients, refining existing empirical formulas. The model's maneuvering …