Computer Engineering Commons^™

Evaluation Of Deep Learning Techniques In Road Sign Recognition, Latafat Abbas Gardashova, Haji Fakhraddin Hajiyev

Chemical Technology, Control and Management

Deep learning has transformed the computer vision field and greatly improved the performance and efficiency of road sign recognition systems. This research compares different deep learning methods, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and hybrid models, in terms of their ability to effectively detect and classify road signs under various conditions. The study compares performance measures such as accuracy, processing speed, and robustness to environmental conditions like low lighting, occlusion, and adverse weather. The results show that CNN-based methods, especially those with transfer learning and ensemble techniques, have better performance in real-time scenarios. Problems like computational …

Retracted: Automated Diagnosis Of Orthopedic Patients With Vertebral Column Disorders Using Advanced Mathematical Modeling, Chen Feng, Zhenhua Sun, Xinheng Dai, Hongli Wen

Iraqi Journal for Computer Science and Mathematics

Orthopedic disorders are multifactorial, making accurate diagnosis a significant challenge. This study introduces a novel method for classifying patients into three categories—normal, disc herniation, and spondylolisthesis—using biomechanical parameters derived from diagnostic datasets. To enhance classification accuracy, two meta-heuristic optimization algorithms—the Zebra Optimization Algorithm (ZOA) and Chaos Game Optimization (CGO)—are integrated with Adaptive Boosting (ADAC) and Light Gradient Boosting Machine (LGBM) classifiers. The experimental results reveal that ZOA significantly improves model performance, particularly in the ADAC classifier. The baseline ADAC model achieved a mean accuracy of 0.916, which increased to 0.952 after optimization with ZOA (referred to as the ADZO model). …

Deep Learning And Texture Analysis For Lung And Colon Cancer Predicting, Mohamed M. Neamah, Laith A. Al-Ani, Loay E. George

Iraqi Journal for Computer Science and Mathematics

Cancer remains a major cause of death worldwide, with lung and colon (LC) cancers presenting significant challenges to healthcare systems due to their high rates of occurrence and mortality. Early and precise diagnosis is essential for better patient outcomes. This research utilizes recent advances in deep learning (DL) and texture analysis (TA) to create a reliable predictive model for detecting LC cancer through histopathological images (HPI). A hybrid method is proposed that combines a gray-level co-occurrence matrix (GLCM) for extracting texture features with an adaptive modified EfficientNet B2 model (AM-EfficientNet B2) for deep feature extraction. These features are used to …

Retracted: Software Engineering-Oriented Text Generation And Analysis Using Gpt-2, Nadia Mahmood Hussien, Aumama Mohammed Farhan, Yasmin Makki Mohialden, Qabas Abdal Zahraa Jabbar, Shahbaa Mohammed Abdulmaged, Asmaa Hatem Arif

Iraqi Journal for Computer Science and Mathematics

The research focuses on developing an improved system for generating and analyzing text by combining GPT-2, LSTM, and CNN models to address challenges in automated content creation for software engineering tasks. The system targets specific applications such as requirements engineering, software documentation, and code comment generation. It generates 150-token text samples based on over 100 user-provided prompts. These generated texts are first processed through an LSTM layer to capture semantic meaning, then passed through a CNN module to extract syntactic and semantic features. All outputs are stored in structured CSV files to support future analysis. Evaluation results demonstrate positive impacts …

Efficient Inference Of Performance Models In Openmp Applications, Gaurav Punjabi

Computer Science and Engineering Master's Theses

Choosing the best OpenMP parameters such as thread count, scheduling type, and chunk size is essential for optimizing parallel program performance. One of the promising approaches is to infer a (pre-trained) performance model at runtime to determine the parameters to run OpenMP parallel regions. Such a performance prediction model can require programs’ code information such as intermediate representation (IR) and other at-runtime information (e.g., input sizes) to make a performance prediction. In such a scenario, extracting or querying the IR information at runtime can create a significant runtime overhead. This thesis proposes a compiler-asssited tuning framework that shifts IR extraction …

Adversaguard: A Distributed Data-Poisoning Benchmark For Parallel Ai, Yulia Kumar, Solomon Thomas, Dejaun Gayle, J. Jenny Li, Dov Kruger

Center for Cybersecurity

As organizations scale model training across large clusters and clouds, data poisoning has emerged as a significant practical threat. Most existing research focuses on data poisoning in single-node environments. Far fewer studies have compared the effectiveness of attacks across parallel training strategies, where factors like gradient aggregation and distributed memory ceilings fundamentally alter attack detectability and its impacts. To address this gap, we introduce AdversaGuard, a reproducible benchmark and accompanying application designed specifically for distributed settings to protect AI training pipelines. This research makes the following key contributions: (1) A comprehensive Distributed Data Poisoning (DDP) benchmark spanning seven distributed systems …

Enhancing Non-Player Character Dialogue In Video Gages: An Evaluation Of Large Language Model-Generated Responses, Lam P. Quach

Master's Theses

As video games increasingly emphasize narrative depth and player immersion, the quality of Non-Player Character (NPC) dialogue has become crucial for creating engaging gaming experiences. This thesis investigates the potential of Large Language Models (LLMs) to generate high-quality NPC dialogue by comprehensively evaluating four state-of-the-art models: Gemma 3 27B, Mistral 7B, QWEN 2.5, and LLAMA 3.1. The study employs a mixed-methods approach, combining human evaluation (N=50 participants) with AI-based assessment across five key benchmarks: coherence, personality expression, engagement, style/tone appropriateness, and overall quality. Participants evaluated 32 dialogue samples (8 per model) generated for a fantasy game context featuring two distinct …

Multi-Modal Depth Estimation Using Camera And Mmwave Sensors, Alston D. Devero-Belfon

Student Theses

For accurately estimating the depth of environments with varying lighting conditions, reliable methods are limited. By utilizing wireless sensor technology in conjunction with cameras, a wide range of environments can be visualized, and objects within these environments can be tracked and monitored. Such methods offer cost-effective alternatives and provide a more secure, data-at-rest option for individuals with low vision, while also enhancing machine perception. In this work, we develop such a prototype that utilizes wireless sensors and cameras, which act in sync, enabling us to estimate the depth of objects within varying lighting environments to a level that is recognizable …

Fpga-Based Overlay Accelerators With Massive Parallel Processing Units To Accelerate Deep Neural Networks, Ehsan Kabir

Graduate Theses and Dissertations

Deep neural networks (DNNs) are widely used in applications such as classification, prediction, and regression. Various DNN architectures, such as convolutional neural networks (CNN), multilayer perceptrons (MLP), long short-term memory (LSTM), recurrent neural networks (RNN), and transformers, have become leading machine learning techniques in these applications. They require significant computational resources and have substantial memory demands due to intensive matrix-matrix multiplications and complex data flows. Hence, efficient utilization of on-chip computational and memory resources is essential to maximize parallelism and minimize latency. Designing an optimal tiling scheme that aligns effectively with the architecture is also necessary. Modern FPGAs are equipped …

Low-Resource Ecoacoustic Audio Classification, Enis Berk Coban

Dissertations, Theses, and Capstone Projects

Ecoacoustic monitoring via machine learning enables scalable analysis but is often constrained by labeled data scarcity, particularly in remote regions like the Arctic. This thesis confronts low-resource ecoacoustic audio classification by developing and evaluating complementary machine learning methodologies. We introduce EDANSA, the first publicly available, expert- labeled Arctic dataset of its kind, curated via novel active learning, alongside a baseline CNN. We systematically evaluate transfer learning, showing general audio embeddings effectively bootstrap classifiers for challenging Arctic sounds, significantly outperforming direct label mapping. Optimizing label utility, we investigate standard data augmentation and introduce novel audio data valuation via Shapley values, revealing …

Text-Based Propaganda Detection, Vivek Sharma

Dissertations, Theses, and Capstone Projects

In the digital age, the swift consumption of news and the widespread dissemination of propaganda significantly threaten democratic principles and informed decision-making. News organizations, often aligned politically, tailor their content to appeal to specific audiences, thereby perpetuating biases. Despite advancements in detecting propaganda through various computational techniques, including Large Language Models(LLM) and BERT-based approaches, significant challenges persist concerning data availability and the effectiveness of detection tools. This thesis explores the landscape of text-based propaganda and reviews related research in textual analysis. It introduces a novel method for collecting a social-media based dataset, which includes analysis of journalists' tweets to discern …

Storage Location Optimization In Automated Storage And Retrieval Systems: A Deep Reinforcement Learning Approach, Lingjun Wang, Aldy Gunawan, Pieter Vansteenwegen

Research Collection School Of Computing and Information Systems

This study investigates the optimization of storage location in automated storage and retrieval systems (AS/RS). We introduce an optimization approach based on the Deep Q-Network (DQN) algorithm to enhance warehouse task efficiency and minimize stacker travel during storage and retrieval. To accelerate the algorithm training process, we integrate a prioritized experience replay mechanism. Furthermore, we decouple action selection from value estimation within the DQN framework to address the issue of value overestimation. The proposed model is evaluated against three heuristic methods. The experimental results demonstrate that our approach significantly outperforms these baselines.

Toward A Generalizable Perceptual Hashing Framework For Image Manipulation Detection, Priyanka Samanta

Dissertations, Theses, and Capstone Projects

This thesis contributes to research in adversarial image manipulation detection. The primary motivation is the increasing need to verify digital images, especially for legal evidence, journalistic proof, or social media content—where manipulated or fabricated images can mislead, defame, or distort reality. A key application and contribution of this work is the development of eWitness, a blockchain application that generates and registers image provenance at capture time to enable independent verification of authenticity. The secret sauce behind the system is SmartHash, a novel and efficient perceptual hashing algorithm designed for real-world deployment in systems like eWitness. Unlike existing algorithms, SmartHash targets …

Enhancing Network Security: Dynamical Intrusion Detection Systems Leveraging Zero Trust Architecture, Ekramul Haque

Tennessee State University Alumni Theses and Dissertations

This thesis discussed two original methodologies for proposing security frameworks incorporating machine learning (ML) and Zero Trust Architecture (ZTA) principles to manage advanced persistent threats faced by Unmanned Aerial Vehicles (UAVs) and Network intrusion detection systems (IDS). The first methodology examined the use of RF signals and deep learning to identify and classify UAVs. The RF signal characteristics used in the method for detecting UAVs improved the ability to determine RF drone protocols. Although the models led to promising findings, their lack of ability to generalize to new drone types identified the need to improve both the data set and …

Optimizing Hip And Knee Assistance For Walking And Sit-To-Stand Transitions: An Intrinsic Muscle Mechanics Based Predictive Approach, Neethan Ratnakumar

Dissertations

As the global population ages, the demand for wearable assistive technologies continues to rise, driven by their potential to enhance mobility and independence in older adults. Effectively designed controllers for lower-limb exoskeletons to assist sit-to-stand (STS) and walking are crucial for delivering efficient, safe, and comfortable assistance during daily activities. Traditionally, controller optimization involves biomechanical modeling and user-specific customization. Musculoskeletal simulations play a central role in this process by providing insights into human-exoskeleton interaction dynamics, thereby informing and refining control strategies.

This work presents a simulation-driven approach for developing exoskeleton controllers for walking and STS using two distinct methods: optimal …

Multimodal Learning In Real-World Application: Enhancing Feature Representation And Training Strategies, Nana Lin

Graduate Doctoral Dissertations

Multimodal learning has emerged as a critical paradigm for developing intelligent systems that can understand and reason across diverse inputs such as images, text, and audio data. Despite significant advances, effective deployment of multimodal models in practice remains a challenging task. This dissertation explores how multimodal learning can be effectively applied to high-stakes, real-world scenarios, with a focus on enhancing feature representation and training efficiency. Specifically, this research investigates multimodal learning strategies in two key domains: healthcare and surveillance.

In the healthcare domain, we explored the data fusion and alignment approaches for cognitive decline diagnoses. First, we propose the LOVEMA …

The Utilisation Of The Fourth Industrial Revolution (4ir) Technologies In E-Government Service Delivery: A Systematic Literature Review, Arnet Zitha, Noluntu Mpekoa, Sheethal Tom

African Conference on Information Systems and Technology

The integration of Fourth Industrial Revolution (4IR) technologies is transforming e-Government by boosting citizen engagement and enhancing service efficiency. However, gaps still exist in understanding the various applications, impacts, and barriers to adoption. This systematic review synthesises literature from 16 studies published between 2017 and 2025, illustrating how technologies like blockchain, artificial intelligence, big data, Internet of Things, and machine learning are employed and their effects on e-Government service delivery. The review reveals that 4IR technologies continue to play a vital role in e-Government services by addressing security threats, simplifying verification and authentication, building trust, and improving the quality and …

Human-Machine Communication: Complete Volume. Volume 10

Human-Machine Communication

This is the complete volume of HMC Volume 10.

The Emergence Of Ai Chatbots In Education, Trek Martin

Journal of Graduate Education Research

The recent advent of popular AI applications in educational contexts has sparked renewed interest in the question of AI and guided learning platforms as teaching tools. What are the possibilities for learning? In the attempt to answer that question, limitations of the field must be brought to full attention, as well an understanding of whether or not those limitations will continue into the immediate future; this research examines the technical evolution of artificial intelligence in education, from early symbolic reasoning systems to modern machine-learning-based chatbots. It then examines that evolution in terms of the key challenges it faced throughout, and …

Smart Mobility Technologies In Urban Areas Of Emerging Economies: A Bibliometric Analysis, Peter Mugisha, Rose Luke, Joash Mageto, Hossana Twinomurinzi

African Conference on Information Systems and Technology

Despite the adoption of smart mobility solutions in emerging economies, challenges such as traffic congestion, pollution and inadequate infrastructure still persist. This study analyses 540 scholarly articles published between 2003 and 2024 to evaluate how smart mobility technologies – such as Intelligent Transportation Systems (ITS), Internet of Things (IoT) and Artificial Intelligence (AI) – have been implemented in these regions. Data was retrieved from Scopus and Web of Science and analysed using Biblioshiny for bibliometric mapping and Atlas.ti for thematic analysis. The review identifies research trends and gaps, showing how ITS has improved transport management in cities like Nairobi, and …