SACM - United States of America
Permanent URI for this collectionhttps://drepo.sdl.edu.sa/handle/20.500.14154/9668
Browse
16 results
Search Results
Item Restricted Next-Generation Diagnostics: Deep Learning based Approaches for Medical Image Analysis(Florida Institute of Technology, 2024-12) Alsubaie, Mohammed; Li, XianqiHigh-resolution medical imaging plays a pivotal role in accurate diagnostics and effective patient care. However, the extended acquisition times required for detailed imaging often lead to patient discomfort, motion artifacts, and increased scan failures. To address these challenges, advanced deep learning approaches are emerging as transformative tools in medical imaging. In this study, we propose a conditional denoising diffusion model-based framework designed to enhance the resolution and reconstruction quality of medical images, including Magnetic Resonance Imaging (MRI) and Magnetic Resonance Spectroscopic Imaging (MRSI). The framework incorporates a data fidelity term into the reverse sampling process to ensure consistency with physical acquisition models while improving reconstruction accuracy. Furthermore, it leverages a Self-Attention UNet architecture to upsample low-resolution MRSI data, preserving fine-grained details and critical structural information essential for clinical diagnostics. The proposed model demonstrates adaptability across varying undersampling rates and spatial resolutions, as a network trained on acceleration factor 8 generalizes effectively to other acceleration factors. Evaluations on publicly available fastMRI datasets and MRSI data highlight significant improvements over state-of-the-art methods, achieving superior metrics in SSIM, PSNR, and LPIPS while maintaining diagnostic relevance. Notably, the diffusion model excels in preserving intricate structural details, detecting small tumors, and maintaining texture integrity, particularly in glioma imaging for mapping tumor metabolism associated with IDH1 and IDH2 mutations. These findings underscore the potential of deep learning-based diffusion models to revolutionize medical imaging, enabling faster, more accurate scans and improving diagnostic workflows across clinical and research applications.12 0Item Restricted Deep Learning Approaches for Multivariate Time Series: Advances in Feature Selection, Classification, and Forecasting(New Mexico State University, 2024) Alshammari, Khaznah Raghyan; Tran, Son; Hamdi, Shah MuhammadIn this work, we present the latest developments and advancements in the machine learning-based prediction and feature selection of multivariate time series (MVTS) data. MVTS data, which involves multiple interrelated time series, presents significant challenges due to its high dimensionality, complex temporal dependencies, and inter-variable relationships. These challenges are critical in domains such as space weather prediction, environmental monitoring, healthcare, sensor networks, and finance. Our research addresses these challenges by developing and implementing advanced machine-learning algorithms specifically designed for MVTS data. We introduce innovative methodologies that focus on three key areas: feature selection, classification, and forecasting. Our contributions include the development of deep learning models, such as Long Short-Term Memory (LSTM) networks and Transformer-based architectures, which are optimized to capture and model complex temporal and inter-parameter dependencies in MVTS data. Additionally, we propose a novel feature selection framework that gradually identifies the most relevant variables, enhancing model interpretability and predictive accuracy. Through extensive experimentation and validation, we demonstrate the superior performance of our approaches compared to existing methods. The results highlight the practical applicability of our solutions, providing valuable tools and insights for researchers and practitioners working with high-dimensional time series data. This work advances the state of the art in MVTS analysis, offering robust methodologies that address both theoretical and practical challenges in this field.14 0Item Restricted TOWARDS ROBUST AND ACCURATE TEXT-TO-CODE GENERATION(University of Central Florida, 2024) almohaimeed, saleh; Wang, LiqiangDatabases play a vital role in today’s digital landscape, enabling effective data storage, manage- ment, and retrieval for businesses and other organizations. However, interacting with databases often requires knowledge of query (e.g., SQL) and analysis, which can be a barrier for many users. In natural language processing, the text-to-code task, which converts natural language text into query and analysis code, bridges this gap by allowing users to access and manipulate data using everyday language. This dissertation investigates different challenges in text-to-code (including text-to-SQL as a subtask), with a focus on four primary contributions to the field. As a solution to the lack of statistical analysis in current text-to-code tasks, we introduce SIGMA, a text-to- Code dataset with statistical analysis, featuring 6000 questions with Python code labels. Baseline models show promising results, indicating that our new task can support both statistical analysis and SQL queries simultaneously. Second, we present Ar-Spider, the first Arabic cross-domain text-to-SQL dataset that addresses multilingual limitations. We have conducted experiments with LGESQL and S2SQL models, enhanced by our Context Similarity Relationship (CSR) approach, which demonstrates competitive performance, reducing the performance gap between the Arabic and English text-to-SQL datasets. Third, we address context-dependent text-to-SQL task, often overlooked by current models. The SParC dataset was explored by utilizing different question rep- resentations and in-context learning prompt engineering techniques. Then, we propose GAT-SQL, an advanced prompt engineering approach that improves both zero-shot and in-context learning experiments. GAT-SQL sets new benchmarks in both SParC and CoSQL datasets. Finally, we introduce Ar-SParC, a context-dependent Arabic text-to-SQL dataset that enables users to interact with the model through a series of interrelated questions. In total, 40 experiments were conducted to investigate this dataset using various prompt engineering techniques, and a novel technique called GAT Corrector was developed, which significantly improved the performance of all base- line models.11 0Item Restricted ADAPTIVE INTRUSION DETECTION SYSTEM FOR THE INTERNET OF MEDICAL THINGS (IOMT): ENHANCING SECURITY THROUGH IMPROVED MUTUAL INFORMATION FEATURE SELECTION AND META-LEARNING(Towson University, 2024-12) Alalhareth, Mousa; Hong, SungchulThe Internet of Medical Things (IoMT) has revolutionized healthcare by enabling continuous patient monitoring and diagnostics but also introduces significant cybersecurity risks. IoMT devices are vulnerable to cyber-attacks that threaten patient data and safety. To address these challenges, Intrusion Detection Systems (IDS) using machine learning algorithms have been introduced. However, the high data dimensionality in IoMT environments often leads to overfitting and reduced detection accuracy. This dissertation presents several methodologies to enhance IDS performance in IoMT. First, the Logistic Redundancy Coefficient Gradual Upweighting Mutual Information Feature Selection (LRGU-MIFS) method is introduced to balance the trade-off between relevance and redundancy, while improving redundancy estimation in cases of data sparsity. This method achieves 95% accuracy, surpassing the 92% reported in related studies. Second, a fuzzy-based self-tuning Long Short-Term Memory (LSTM) IDS model is proposed, which dynamically adjusts training epochs and uses early stopping to prevent overfitting and underfitting. This model achieves 97% accuracy, a 10% false positive rate, and a 94% detection rate, outperforming prior models that reported 95% accuracy, a 12% false positive rate, and a 93% detection rate. Finally, a performance-driven meta-learning technique for ensemble learning is introduced. This technique dynamically adjusts classifier voting weights based on factors such as accuracy, loss, and prediction confidence levels. As a result, this method achieves 98% accuracy, a 97% detection rate, and a 99% F1 score, while reducing the false positive rate to 10%, surpassing previous results of 97% accuracy, a 93% detection rate, a 97% F1 score, and an 11% false positive rate. These contributions significantly enhance IDS effectiveness in IoMT, providing stronger protection for sensitive medical data and improving the security and reliability of healthcare networks.10 0Item Restricted DEEP LEARNING FOR MOLECULAR DESIGN: MODELS, FRAMEWORKS, AND APPLICATIONS(Cornell University, 2024-08) Alshehri, Abdulelah Saeed; You, Fengqi; Gomes, Carla; Abbott, Nicholas L.The vast and complex landscape of chemical space has traditionally been explored through a combination of experimentation and knowledge-based computational approaches. However, the limitations of these methods have hindered the efficient design of molecules with desired properties. The advent of deep learning, coupled with the availability of big chemical data, presents transformative opportunities for computational molecular design. This dissertation explores the convergence of deep learning and chemical engineering, presenting novel methodologies and frameworks to address challenges in molecular property prediction, molecular design, chemical data extraction, molecular conformation generation, and peptide design. In Chapter 2, we develop parallel models for the estimation of 25 pure component properties across over 24,000 chemicals, employing both traditional regression and machine learning methods on functional group representations. These models demonstrate robust accuracy in predicting a broad range of physicochemical properties, enabling streamlined product and process design. Chapter 3 addresses the inherent uncertainty in CMD by introducing DRL-CMD, an uncertainty-aware deep reinforcement learning framework. By explicitly quantifying and managing uncertainties, DRL-CMD reduces constraint violations by 39% and uncertainty margins by 27% compared to literature-reported molecules, particularly in complex design scenarios with limited data and extreme property ranges. This approach offers a more reliable path to molecules with tailored properties toward accelerating product and process design. In Chapter 4, the focus is on the extraction of chemical data from scientific literature, critical for model training and discovery. ChemREL, a novel deep learning pipeline, achieves an F1-score of 95.4% for property extraction, outperforming existing methods and GPT-4. Its transferability is demonstrated by successful adaptation from melting point extraction to LD50 extraction with minimal additional training, highlighting the potential to accelerate the construction of large-scale chemical datasets. In Chapter 5, we explore the utilization of abundant 2D molecular graph data to enhance 3D conformer generation, a crucial step in drug discovery. By pretraining graph neural networks on 2D data and improving the GeoMol method, we achieve a 7.7% average improvement in generated conformer quality compared to state-of-the-art sequential methods, improving the accuracy and efficiency of molecular modeling. Chapter 6 addresses the global challenge of plastic pollution by presenting an integrated framework combining biophysics-based insights, evidential deep learning, and metaheuristic search for the design of plastic-binding peptides. This approach leads to significant increases in binding free energies for polypropylene (18%) and polystyrene (34%) compared to previous designs, offering a promising bio-inspired solution for plastic remediation. By developing these novel deep learning approaches, the resulting advances improve predicting molecular properties, designing molecules with tailored properties while managing uncertainties, constructing a versatile pipeline for chemical data extraction, enhancing the quality of 3D conformer generation, and generating high-affinity plastic-binding peptides for potential environmental remediation. These works signify a step forward in the integration of deep learning and chemical engineering, paving the way for accelerated discovery and innovation in the field.23 0Item Restricted Multi-Stage and Multi-Target Data-Centric Approaches to Object Detection, Localization, and Segmentation in Medical Imaging(University of California San Diego, 2024) Albattal, Abdullah; Nguyen, TruongObject detection, localization, and segmentation in medical images are essential in several medical procedures. Identifying abnormalities and anatomical structures of interest within these images remains challenging due to the variability in patient anatomy, imaging conditions, and the inherent complexities of biological structures. To address these challenges, we propose a set of frameworks for real-time object detection and tracking in ultrasound scans and two frameworks for liver lesion detection and segmentation in single and multi-phase computed tomography (CT) scans. The first framework for ultrasound object detection and tracking uses a segmentation model weakly trained on bounding box labels as the backbone architecture. The framework outperformed state-of-the-art object detection models in detecting the Vagus nerve within scans of the neck. To improve the detection and localization accuracy of the backbone network, we propose a multi-path decoder UNet. Its detection performance is on par with, or slightly better than, the more computationally expensive UNet++, which has 20% more parameters and requires twice the inference time. For liver lesion segmentation and detection in multi-phase CT scans, we propose an approach to first align the liver using liver segmentation masks followed by deformable registration with the VoxelMorph model. We also propose a learning-free framework to estimate and correct abnormal deformations in deformable image registration models. The first framework for liver lesion segmentation is a multi-stage framework designed to incorporate models trained on each of the phases individually in addition to the model trained on all the phases together. The framework uses a segmentation refinement and correction model that combines these models' predictions with the CT image to improve the overall lesion segmentation. The framework improves the subject-wise segmentation performance by 1.6% while reducing performance variability across subjects by 8% and the instances of segmentation failure by 50%. In the second framework, we propose a liver lesion mask selection algorithm that compares the separation of intensity features between the lesion and surrounding tissue from multi-specialized model predictions and selects the mask that maximizes this separation. The selection approach improves the detection rates for small lesions by 15.5% and by 4.3% for lesions overall.19 0Item Restricted Towards Cost-Effective Noise-Resilient Machine Learning Solutions(University of Georgia, 2026-06-04) Gharawi, Abdulrahman Ahmed; Ramaswamy, LakshmishMachine learning models have demonstrated exceptional performance in various applications as a result of the emergence of large labeled datasets. Although there are many available datasets, acquiring high-quality labeled datasets is challenging since it involves huge human supervision or expert annotation, which are extremely labor-intensive and time-consuming. The problem is magnified by the considerable amount of label noise present in datasets from real-world scenarios, which significantly undermines the performance accuracy of machine learning models. Since noisy datasets can affect the performance of machine learning models, acquiring high-quality datasets without label noise becomes a critical problem. However, it is challenging to significantly decrease label noise in real-world datasets without hiring expensive expert annotators. Based on extensive testing and research, this dissertation examines the impact of different levels of label noise on the accuracy of machine learning models. It also investigates ways to cut labeling expenses without sacrificing required accuracy. Finally, to enhance the robustness of machine learning models and mitigate the pervasive issue of label noise, we present a novel, cost-effective approach called Self Enhanced Supervised Training (SEST).21 0Item Restricted A Deep Learning Framework for Blockage Mitigation in mmWave Wireless(Portland State University, 2024-05-28) Almutairi, Ahmed; Aryafar, EhsanMillimeter-Wave (mmWave) communication is a key technology to enable next generation wireless systems. However, mmWave systems are highly susceptible to blockages, which can lead to a substantial decrease in signal strength at the receiver. Identifying blockages and mitigating them is thus a key challenge to achieve next generation wireless technology goals, such as enhanced mobile broadband (eMBB) and Ultra-Reliable and Low-Latency Communication (URLLC). This thesis proposes several deep learning (DL) frameworks for mmWave wireless blockage detection, mitigation, and duration prediction. First, we propose a DL framework to address the problem of identifying whether the mmWave wireless channel between two devices (e.g., a base station and a client device) is Lineof- Sight (LoS) or non-Line-of-Sight (nLoS). Specifically, we show that existing beamforming training messages that are exchanged periodically between mmWave wireless devices can also be used in a DL model to solve the channel classification problem with no additional overhead. We extend this DL framework by developing a transfer learning model (t-LNCC) that is trained on simulated data and can successfully solve the channel classification problem on any commercial-off-the-shelf (COTS) mmWave device with/without any real-world labeled data. The second part of the thesis leverages our channel classification mechanism from the first part and introduces new DL frameworks to mitigate the negative impacts of blockages. Previous research on blockage mitigation has introduced several model and protocol based blockage mitigation solutions that focus on one technique at a time, such as handoff to a different base station or beam adaptation to the same base station. We go beyond those techniques by proposing DL frameworks that address the overarching problem: what blockage mitigation method should be employed? and what is the optimal sub-selection within that method? To do so, we developed two Gated Recurrent Unit (GRU) models that are trained using periodically exchanged messages in mmWave systems. Specifically, we first developed a GRU model that tackled the blockage mitigation problem in single-antenna clients wireless environment. Then, we proposed another GRU model to expand our investigation to cover more complex scenarios where both base stations and clients are equipped with multiple antennas and collaboratively mitigate blockages. Those two models are trained on datasets that are gathered using a commercially available mmWave simulator. Both models achieve outstanding results in selecting the optimal blockage mitigation method with an accuracy higher than 93% and 91% for single-antenna and multiple-antenna clients, respectively. We also show that the proposed methods significantly increases the amount of transferred data compared to several other blockage mitigation policies.17 0Item Restricted ENSEMBLE MACHINE LEARNING IN SPACE WEATHER ANALYTICS(New Jersey Institute of Technology, 2024) Alobaid, Khalid; Wang, JasonThis dissertation addresses several important space weather problems using ensemble learning techniques. An ensemble method combining multiple machine learning models is often more accurate than the individual machine learning models that form the ensemble method. There are several techniques for constructing an ensemble. With in-depth case studies, the dissertation demonstrates the usefulness and effectiveness of ensemble machine learning for space weather analytics, especially for predicting extreme space weather events such as coronal mass ejections (CMEs). The dissertation begins with an ensemble method for predicting the arrival time of CMEs from the Sun to Earth. The proposed method, named CMETNet, combines classical machine learning algorithms such as support vector regression, random forests, XGBoost and Gaussian process regression, along with a deep convolutional neural network (CNN), to perform multimodal learning. The classical machine learning algorithms are used to learn latent patterns from CME features and background solar wind parameters while the deep CNN is used to learn patterns hidden in CME images where the learned patterns are jointly used to make predictions. Experimental results show that CMETNet outperforms existing models, both machine learning based and physics based. Finally, the dissertation presents a fusion method, named DeepCME, to estimate two important properties of CMEs, namely, CME mass and kinetic energy. The DeepCME method is a fusion of three deep-learning models, namely ResNet, InceptionNet, and InceptionResNet. The fusion model extracts features from Large Angle and Spectrometric Coronagraph (LASCO) C2 images, effectively combining the learning capabilities of the three component models to jointly estimate the mass and kinetic energy of CMEs. To the best of current knowledge, this is the first time that deep learning has been used for CME mass and kinetic energy estimations. DeepCME can help scientists better understand CME dynamics. In conclusion, the dissertation showcases many applications of learning techniques including ensemble learning, deep learning, transfer learning and multimodal learning in space weather analytics. The tools and methods developed from the dissertation will make contributions to the understanding and forecasting of CME dynamics and CME geoeffectiveness.16 0Item Restricted MACHINE LEARNING FOR TRAFFIC PREDICTION AND COMMUNICATION EFFICIENT DATA ANALYTIC IN WIRELESS NETWORKS(Georgia Institute of Technology, 2024-05-02) Alamoudi, Abdulrahman; Fekri, FaramarzWith the exponential growth of available data, deep learning has emerged as a fundamental tool for interpreting data abstractions and constructing computational models. It has revolutionized our understanding of information processing, facilitating exploration across diverse domains such as text and signal analysis, image and audio recognition, social network analysis, and bioinformatics. The overarching goal of this research is to minimize wireless network traffic and operational costs for mobile users and network operators, respectively. The integrated framework endeavors to develop predictive models by analyzing the behavior of mobile users within wireless networks and designing efficient, task-oriented models in wireless networks. Particularly, our research taps into machine learning to learn and forecast mobile user behaviors, design semantic communication systems over noisy channels, and implement unsupervised distributed functional compression over wireless channels.63 0