Engineering Commons^™

Lifelong Machine Learning With Adaptive Resonance Theory, Sasha Petrenko

Doctoral Dissertations

"This publication option dissertation is composed of three papers concerning the study of the problem lifelong machine learning with Adaptive Resonance Theory (ART) algorithms. Lifelong learning (L2) is a challenging machine learning paradigm that both encompasses and formalizes the fields of continual learning and incremental learning. The field is concerned with the mitigation of the phenomenon of catastrophic forgetting whereby learning agents that are faced with incrementally novel information deleteriously overwrite previous knowledge if that learning process is not regularized to counteract this consequence. ART algorithms solve this stability-plasticity dilemma by optimally assigning learning to categories or instantiating new knowledge …

Transplant Surgeon Fuzzy Associative Memory (Tsfam): Model For Capturing Surgeon Perspective, Rachel Dzieran, Cihan H. Dagli, Robert J. Marley

Engineering Management and Systems Engineering Faculty Research & Creative Works

AI-driven healthcare decision-making is multi-faceted, requiring complex logic to adapt to evolving policies and societal demands. Effective change implementation by healthcare providers and multidisciplinary organ transplant teams depends on adaptive decision-making. The proposed Transplant Surgeon Fuzzy Associative Memory (TSFAM) model introduces a novel approach to Human-AI Teaming, keeping human expertise central while dynamically adjusting to changing requirements. TSFAM employs fuzzy logic to manage imperfect data and human ambiguity, integrating the transplant surgeon perspective with the AI deep learning decision-making tool, creating a resilient solution in this critical domain. By embedding adaptive capabilities into the architecture, TSFAM exemplifies the adaptability of …

A Cost-Effective Nilm Solution With Three-Point Labelling And Non-Causal Convolution Technique, Yanan Zhang, Gan Zhou, Yanjun Feng, Zhan Liu, Li Huang, Zhi Li, Rui Bo

Electrical and Computer Engineering Faculty Research & Creative Works

Although deep learning is increasingly promising in the field of Non-Intrusive Load Monitoring (NILM) these days, the high costs of data recording and labelling represent a significant challenge for the training of supervised models. To address this, a cost-effective sequence-to-points NILM solution is proposed, integrating three-point labelling with non-causal convolution techniques. The approach introduces a semi-automatic labelling framework for obtaining NILM three-point data, which provides a low-cost data collection and labelling solution for large-scale applications. Then, a novel loss function combining coordinate loss and confidence loss is developed to address the positional misalignment and negative sample confusion in sequence-to-points scenario …

Smart: Semantic Mapping And Analysis For Regional Terrain Using Multi-Scale U-Net And Topsis, Rashmi Sonth

Master's Projects

Accurate land use classification is the backbone for urban planning. But with poor quality satellite images, varied landscapes and structures which are changing faster than ever, it becomes a challenge to define clear boundaries and hence to urban planning. This research explores the application of deep-learning model for land use classification and asses the suitability of the land. The proposed model combines a multi-scale U-Net architecture with Transformer blocks applied on a multi-spectral satellite images that improves the semantic segmentation greatly across the urban and rural regions. Additionally, a patch-wise segmentation is applied to overcome the common problem of feature …

Ai-Powered Image-Based Assessment Of Pressure Injuries Using You Only Look Once (Yolo) Version 8 Models, Mehedi Hasan Tusar, Fateme Fayyazbakhsh, Niloofar Zendehdel, Eduard Mochalin, Igor Melnychuk, Lisa Gould, Ming C. Leu

Mechanical and Aerospace Engineering Faculty Research & Creative Works

Objective: The primary objective of this study is to enhance the detection and staging of pressure injuries using machine learning capabilities for precise image analysis. This study explores the application of the You Only Look Once version 8 (YOLOv8) deep learning model for pressure injury staging. Approach: We prepared a high-quality, publicly available dataset to evaluate different variants of YOLOv8 (YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l, and YOLOv8x) and five optimizers (Adam, AdamW, NAdam, RAdam, and stochastic gradient descent) to determine the most effective configuration. We followed a simulation-based research approach, which is an extension of the Consolidated Standards of Reporting Trials …

Cnns And Transformers For Visual Understanding: From Feature Alignment To Image Captioning, Xuehao Liu

Doctoral

Deep learning has developed rapidly since the introduction of Deep Belief Networks during the past decade. As an area of machine learning, it still has many open challenges. Among these open challenges is the issue of transparency, with deep learning models known as black boxes. Both explainability of a model for understanding the decision making process, and the transparency of the relationship between the input and output are crucial for understanding a model. The understanding of models can build trust between AI systems and humans, verify models behavior, and identify potential biases or errors.

Deep Learning For Computer Vision Applications In Medical Diagnostics And Wildlife Monitoring, Mostapha Al Saidi

Electronic Theses and Dissertations

This dissertation explores innovative applications of deep learning and computer vision techniques across three distinct domains: medical imaging, dermatological diagnostics, and wildlife monitoring. The research addresses critical challenges in each field through the development and optimization of convolutional neural networks and other deep learning architectures.

The first study examines COVID-19 classification from X-ray images, comparing one-shot versus two-stage classification approaches using transfer learning with pre-trained models such as VGG16 and VGG19. The initial hypothesis was that breaking down the classification task into two optimized tasks would yield better results than one-shot classification. Results demonstrated that the single-stage approach achieved superior …

Learning From Leads: A 1d Dilated Resnet For Ecg Chagas Disease Screening, Somesh Saini, Matheus Lima Diniz Araujo

Student Scholarship

No abstract provided.

Long-Term Traffic Prediction Using Deep Learning Long Short-Term Memory, Ange-Lionel Toba, Sameer Kulkarni, Wael Khallouli, Timothy Pennington

School of Cybersecurity Faculty Publications

Traffic conditions are a key factor in our society, contributing to quality of life and the economy, as well as access to professional, educational, and health resources. This emphasizes the need for a reliable road network to facilitate traffic fluidity across the nation and improve mobility. Reaching these characteristics demands good traffic volume prediction methods, not only in the short term but also in the long term, which helps design transportation strategies and road planning. However, most of the research has focused on short-term prediction, applied mostly to short-trip distances, while effective long-term forecasting, which has become a challenging issue …

Development Of An Ecg-Based Deep Learning Model For Pediatric Congenital Heart Disease (Chd) Diagnosis, Annbar Mekouar

Selected Full-Text Master Theses 2021-

Congenital heart disease (CHD) stands as the leading congenital anomaly which affects pediatric populations throughout the world. The effectiveness of treatment depends on both early diagnosis and accurate identification but echocardiography requires manual interpretation which proves time-consuming and inconsistent especially when examining pediatric patients with their distinct cardiac systems. The research aims to create a deep learning-based diagnostic framework which uses ECG data to identify coronary artery disease subtypes in pediatric patients. The model uses high-quality datasets from Dr. Ignacio Lugones to extract R-R intervals and QRS durations through convolutional neural networks (CNNs). The system addresses pediatric-specific challenges while enhancing …

Physics-Informed Deep Learning With Kalman Filter Mixture For Traffic State Prediction, Niharika Deshpande, Hyoshin (John) Park

Engineering Management & Systems Engineering Faculty Publications

Accurate traffic forecasting is crucial for understanding and managing congestion for efficient transportation planning. However, conventional approaches often neglect epistemic uncertainty, which arises from incomplete knowledge across different spatiotemporal scales. This study addresses this challenge by introducing a novel methodology to establish dynamic spatiotemporal correlations that captures the unobserved heterogeneity in travel time through distinct peaks in probability density functions, guided by physics-based principles. We propose an innovative approach to modifying both prediction and correction steps of the Kalman Filter (KF) algorithm by leveraging established spatiotemporal correlations. Central to our approach is the development of a novel deep learning model …

Leveraging Transformer-Based Ocr Model With Generative Data Augmentation For Engineering Document Recognition, Wael Khallouli, Mohammad Shahab Uddin, Andres Sousa-Poza, Jiang Li, Samuel Kovacic

Engineering Management & Systems Engineering Faculty Publications

The long-standing practice of document-based engineering has resulted in the accumulation of a large number of engineering documents across various industries. Engineering documents, such as 2D drawings, continue to play a significant role in exchanging information and sharing knowledge across multiple engineering processes. However, these documents are often stored in non-digitized formats, such as paper and portable document format (PDF) files, making automation difficult. As digital engineering transforms processes in many industries, digitizing engineering documents presents a crucial challenge that requires advanced methods. This research addresses the problem of automatically extracting textual content from non-digitized legacy engineering documents. We introduced …

Cyber Forensics With Deep Learning Recurrent Neural Networks, Pfautch Ric, Dagli Cihan, Ashiku Lirim

Engineering Management and Systems Engineering Faculty Research & Creative Works

Detection of anomalies and anti-patterns is essential for adaptive systems with the ability to perform without foreknowledge. Some problems require both classification and regression along with sensitivity tuning and explainability. Some have highly dimensional datasets that are time dependent. This research offers results for Long-Short-Term Memory (LSTM) and Gated Recurrent Units (GRU) algorithms using the BETH dataset. It unpacks metadata attributes and stages a unique approach via Abstract-Feature Analysis (AFA), hyper parameter tuning, and Principal Component Analysis (PCA) within the RNN model. By removing foreknowledge, this research offers insights into RNN anomaly detection performance when an event absent in training …

Augmenting Missing Sensor Data For Robust Human Activity Recognition, Suryakangeyan Kandasamy Gowdaman

Master's Projects

Applications of ubiquitous computing, including health monitoring, sports analytics, and ambient-assisted living, rely on Human Activity Recognition (HAR) using wearable sensors. However, model robustness is challenged by missing sensor values, class imbalance, inter-subject variability, and temporal noise. This work proposes a complete HAR pipeline that addresses these challenges through sampling, time-series augmentation, dynamic feature handling, and GAN-PCA-based imputation. Built on the DeepSense architecture, the model integrates convolutional feature extraction with bi-GRUs for temporal modeling. The system is evaluated using 5-fold cross-validation, subject-aware holdout, and LOSEO strategies on the Opportunity dataset. Results demonstrate consistent accuracy across folds and strong generalization to …

Suicidal Ideation Detection On Reddit Using Llm-Annotated Data And Graph Neural Networks, Ikbal Singh Gurdev Singh Dhanjal

Master's Projects

Suicide is the fourth leading cause of death among people aged 15-29. More than 720, 000 people commit suicide every year. During the COVID-19 pandemic, we saw an increase in people seeking out mental health support on anonymous forums like Reddit. These anonymous forums allow people to express their suicidal ideation without judgment and give them a support structure that not everyone has. The aim of this project is to detect suicidal ideation using Reddit. In this work, we propose SIRGEL (Suicidal Ideation on Reddit using Graph Embeddings and LLMs), a dual-pipeline approach that combines large language model (LLM)- based …

Improving The Robustness Of Compressed Deep Learning Models Against Class Imbalance, Baraa Saeed Ali

Wayne State University Dissertations

Deep Learning (DL) models are deployed ubiquitously, as they power a wide range of critical applications, including image classification, fraud detection, autonomous vehicles, robots, and NLP. However, their massive size and huge memory footprint (overparameterization) represent a serious challenge to the efficient deployment of such models, especially in resource-scarce environments such as wearable devices, smartphones, edge devices, and embedded systems. Therefore, model compression techniques are typically used to shrink the model size to the currently available computational and memory budget and to accelerate training and inference without sacrificing model accuracy and performance. Therefore, the DL research community considers model compression …

Coral Vision – Crustose Coralline Algae Detection With Computer Vision, Ryan Tseng

Master's Projects

Crustose coralline algae (CCA) are a group of red algae that are vital contributors to the health of coral reef ecosystems. Monitoring CCA abundance can serve as an indicator for coral reef health and improve reef conservation efforts. Autonomous Reef Monitoring Structures (ARMS) are artificial structures that can be deployed into coral reef ecosystems and retrieved to gather ecological data without harming reef structures. Traditional methods of calculating CCA abundance require manual analysis and are labor-intensive. Recent developments in computer vision and deep learning technology have provided an avenue to fully automate this task. This research aims to train a …

Deep Learning-Based Model For Automated Prediction Of Coastal Changes: A Robust Approach To Environmental Forecasting, Tsair-Fwu Lee, Chu-Ho Chang, Chin-Shiuh Shieh, Chih-Hsien Wu, Jen-Chung Shao, Chien-Liang Chiu

Journal of Marine Science and Technology–Taiwan

The coastline stands as a critical domain encompassing industry and the environment. The escalating global warming, leading to elevated sea levels and intensified wave-current interactions, has given rise to substantial coastal erosion. This predicament, in conjunction with excessive human development and Taiwan's coastal areas' extreme climatic impact, accentuates the perceptibility of coastal beach alterations. As a result, coastal erosion has emerged as a pressing issue necessitating resolution. Traditional methodologies for assessing coastline changes have conventionally relied on manual measurements. However, owing to the fluctuating distance of coastlines, influenced by tidal patterns, extended measurement processes over several months are susceptible to …

Implementation Of A Neural Network Execution Framework For Generalized And Cross-Platform Deep Learning Deployment And Inference On Spacecraft Systems, Rafael Polanco Segovia

Graduate Theses, Dissertations, and Problem Reports (ETD)

Recent advances in hardware and software technology have made it possible to implement more resource-demanding deep learning algorithms in constrained hardware environments. This creates opportunities to use deep learning for aerospace applications on increasingly smaller aerospace vehicles. This work presents the implementation of a Neural Network Execution Framework (NNEF), which aims to provide a cross-platform and reusable framework to deploy and execute trained neural networks for deep learning aerospace applications. The NNEF executes any neural network inference process regardless of the original deep learning framework in which it was created, for supported flight software platforms, and space-like computer boards. Users …

The Role Of Artificial Intelligence In Boosting Cybersecurity And Trusted Embedded Systems Performance: A Systematic Review On Current And Future Trends, Xiangyi Cheng, Ahmed Oun, Kaden Wince

Mechanical Engineering Faculty Works

As technology becomes increasingly interconnected, ensuring the security of cyber and embedded systems is critical due to escalating vulnerabilities and sophisticated cyber threats. Researchers are exploring artificial intelligence (AI) to improve security mechanisms, yet there is a lack of a comprehensive technical, AI-focused analysis detailing the integration of AI into existing security hardware and frameworks. To address this gap, this article systematically reviews 63 articles on AI in cybersecurity and trusted embedded systems. The reviewed articles are categorized into four application domains: 1) Intrusion Detection and Prevention (IDPS), 2) Malware Detection, 3) Industrial Control and Cyber-Physical Systems (CPS) and 4) …

A Hybrid Data Processing, Computational Intelligence, And Complex Systems Modeling Approach For Describing And Predicting The Bitcoin Market, Oluwadamilare Akinpelu Omole

Doctoral Dissertations

"The Bitcoin market, like traditional financial markets, is a complex system with intricate interdependencies and nonlinear interactions among various market elements. This, coupled with the inherent uncertainty and high volatility present, makes predicting Bitcoin price movements difficult. The lack of understanding of the underlying market dynamics often results in significant losses for investors and traders. Existing studies have focused on the use of predictive models, which have not sufficiently captured the complexities of the market and cannot forecast extreme market events. This research endeavors to bridge this gap by combining data processing techniques, computational intelligence, and complex systems theory to …

Ad2c-Sg-Tl: Alzheimer's Disease Detection And Classification Based On Stacked Generalization And Transfer Learning Approaches, Mariam Gamal Alboghdady, Amira Y. Haikal, Hesham H. Gad, Noha A. Sakr

Mansoura Engineering Journal

Alzheimer's disease (AD) is a degenerative neurologic illness that causes brain atrophy and cell death. Although there is no cure for AD, diagnosing its onset can be very beneficial in the medical field.This paper presents a deep ensemble learning framework for classifying AD stages. Transfer learning (TL) is applied using eight pretrained convolutional neural networks (CNNs) (i.e., VGG16, VGG19, ResNet50V2, MobileNet, DenseNet121, DenseNet169, Xception and MobileNetV2). Stackedgeneralization ensembles techniques are used to provide greater generalization by combining finetuned models with five ensemble models. Using five different stacked ensembles (SE) models to improve the generalization.The ensemble model created by combining all …

Optimizing Ai Language Models: A Study Of Chatgpt-4 Vs. Chatgpt-4o, Md Nurul Absar Siddiky, Muhammad Enayetur Rahman, Md Fayaz Bin Hossen, Muhammad Rezaur Rahman, Md. Shahadat Jaman

Electrical & Computer Engineering Faculty Publications

This paper presents a comparative analysis of OpenAI's GPT-4 and its optimized variant, GPT-4o, focusing on their architectural differences, performance, and real-world applications. GPT-4, built upon the Transformer architecture, has set new standards in natural language processing (NLP) with its capacity to generate coherent and contextually relevant text across a wide range of tasks. However, its computational demands, requiring substantial hardware resources, make it less accessible for smaller organizations and real-time applications. In contrast, GPT-4o addresses these challenges by incorporating optimizations such as model compression, parameter pruning, and memory-efficient computation, allowing it to deliver similar performance with significantly lower computational …

A Fast Framework For Generating Radioactive Mixture Spectra And Its Application To Remote High-Performance Mixture Identification, Chiman Kwan, Bulent Ayhan, Adam Stavola, Kazi Aminul Islam, Hongfang Zhang, Jiang Li

Electrical & Computer Engineering Faculty Publications

Remote detection of radioactive materials in mixtures using handheld or portal detectors remains a challenge because of factors such as low concentration, environmental interference, sensor noise, and other complications. This work introduces a fast framework for generating realistic mixture spectra. Moreover, we present mixture isotope identification using data generated by the fast framework. Researchers have examined a range of conventional and recent algorithms within the fields of machine learning and deep learning. An application to uranium enrichment-level prediction has been included. Extensive simulation experiments validated the efficacy of the proposed framework.

Adaptive Fusion Neural Networks For Sparse-Angle X-Ray 3d Reconstruction, Shaoyong Hong, Bo Yang, Yan Chen, Hao Quan, Shan Liu, Minyi Tang, Jiawei Tian

Electrical & Computer Engineering Faculty Publications

3D medical image reconstruction has significantly enhanced diagnostic accuracy, yet the reliance on densely sampled projection data remains a major limitation in clinical practice. Sparse-angle X-ray imaging, though safer and faster, poses challenges for accurate volumetric reconstruction due to limited spatial information. This study proposes a 3D reconstruction neural network based on adaptive weight fusion (AdapFusionNet) to achieve high-quality 3D medical image reconstruction from sparse-angle X-ray images. To address the issue of spatial inconsistency in multi-angle image reconstruction, an innovative adaptive fusion module was designed to score initial reconstruction results during the inference stage and perform weighted fusion, thereby improving …

Ai-Based Steganography Method To Enhance The Information Security Of Hidden Messages In Digital Images, Nhi Do Ngoc Huynh, Jiajun Jiang, Chung-Hao Chen, Wen-Chao Yang

Electrical & Computer Engineering Faculty Publications

With the increasing sophistication of Artificial Intelligence (AI), traditional digital steganography methods face a growing risk of being detected and compromised. Adversarial attacks, in particular, pose a significant threat to the security and robustness of hidden information. To address these challenges, this paper proposes a novel AI-based steganography framework designed to enhance the security of concealed messages within digital images. Our approach introduces a multi-stage embedding process that utilizes a sequence of encoder models, including a base encoder, a residual encoder, and a dense encoder, to create a more complex and secure hiding environment. To further improve robustness, we integrate …

Land Target Detection Algorithm In Remote Sensing Images Based On Deep Learning, Wenyi Hu, Xiaomeng Jiang, Jiawei Tian, Shitong Ye, Shan Liu

Electrical & Computer Engineering Faculty Publications

Remote sensing technology plays a crucial role across various sectors, such as meteorological monitoring, city planning, and natural resource exploration. A critical aspect of remote sensing image analysis is land target detection, which involves identifying and classifying land-based objects within satellite or aerial imagery. However, despite advancements in both traditional detection methods and deep-learning-based approaches, detecting land targets remains challenging, especially when dealing with small and rotated objects that are difficult to distinguish. To address these challenges, this study introduces an enhanced model, YOLOv5s-CACSD, which builds upon the YOLOv5s framework. Our model integrates the channel attention (CA) mechanism, CARAFE, and …

Computer Vision-Based Framework For Data Extraction From Heterogeneous Financial Tables: A Comprehensive Approach To Unlocking Financial Insights, Iftakhar Ali Khandokar, Priya Deshpande

Electrical and Computer Engineering Faculty Research and Publications

Information extraction from financial document images is crucial in computer vision and NLP, as financial data often exists in image or PDF format, enabling organizations to analyze and make informed business decisions using OCR advancements. The table contents of financial document images are one of the prominent structures to confine important portions of data of the document and many Deep learning-based methods have been proposed to detect Table regions inside document images. The shortcomings of the current approach are that it is bounded within the detection of the table region and struggles in cases such as handling different layouts and …

A Predictive Iot And Cloud Framework For Smart Healthcare Monitoring Using Integrated Deep Learning Model, Preeti Rani, Umesh Chandra Garjola, Haider Abbas

NJF Intelligent Engineering Journal

The researchers developed a deep learning-based smart healthcare monitoring system based on IoT and cloud technology. The proposed system integrates IoT sensors for real-time collection of physiological data, such as ECG, blood pressure, and heart rate, with cloud computing for secure storage and advanced analytics. Utilizing the Bi-LSTM model with fuzzy inference systems (FIS), the framework enhances the accuracy and efficiency of heart disease prediction. According to the evaluation, the model performs better in terms of accuracy, precision, recall, and F1 score than traditional LSTM and FLSTM models. By enabling early detection and personalized interventions, the system aims to reduce …

Automated Segmentation Of The Ulnar Nerve In Mri Using Deep Learning Techniques, Akhil Nagulapalli

Theses

Cubital Tunnel Syndrome (CuTS), a condition caused by compression of the ulnar nerve, results in numbness, tingling, pain, and even muscle atrophy, affecting fine motor skills and diminishing patient quality of life. Accurate diagnosis of CuTS is challenging, as current diagnostic methods—including clinical exams, nerve conduction studies, and unaided MRI—often lack the precision to reliably identify the nerve and detect compression in its early stages. Deep learning-based segmentation offers a promising solution, enabling precise and automated identification of nerve structures in MRI images, which could significantly improve diagnostic accuracy and support timely intervention.

A novel deep learning model for segmenting …