Computer Sciences Commons^™

Deep Learning Approaches For Voltammetric Analysis Of Coffee, Ryan Koes

Honors Theses

This thesis investigates deep learning approaches for voltammetric analysis of brewed coffee using a low-cost electrochemical system and screen-printed electrodes (SPEs). Traditional analytical methods, such as high-performance liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC-MS), provide precise quantification of key compounds but require expensive instrumentation and specialized expertise, limiting accessibility. While SPEs offer a more accessible alternative, they yielded poor results with traditional processing; however, when combined with a neural network, the system proved more effective. In experiments with 132 coffee samples, mean errors for caffeine, CGA, and TDS predictions were 52.98 ppm, 70.48 ppm, and 0.08%, respectively. These findings …

Dataless Neural Networks For Boolean Satisfiability And Network Optimization, Andrew Evan Gautier

Graduate Theses, Dissertations, and Problem Reports (ETD)

Combinatorial optimization problems (COPs) require searching over a finite solution space subject to constraints, with the goal of satisfying an objective function. They arise in operations research, scheduling, resource allocation, circuit design, and many other fields. Many problems in combinatorial optimization (including satisfiability and network design) are NP-hard. Traditionally, researchers have built approximate solvers that return near- optimal solutions efficiently by developing increasingly sophisticated heuristics and meta- heuristics. Deep learning has provided new opportunities for improving combinatorial solvers by leveraging neural guidance to prune the search space. Traditional neural networks have distinct drawbacks in this context: separate training and …

Multimodal Emotion Detection System, Shubhankar Sameer Munshi

Master's Projects

Trying to understand emotion from speech is a problem that is present in human computer interaction. Nevertheless, there are still some shortcomings in current SER methods. Text-based systems may miss vital vocal cues, such as sarcasm, tone changes, and delivery. On the other hand, purely audio-based systems are prone to noise and unstable acoustic features. The combination of linguistic and acoustic features in multimodal approaches partially solves this problem, but many existing approaches use inflexible multimodal fusion techniques that cannot adjust their behaviors according to the quality of input signals. In this work, we propose a multimodal approach based on …

On-Device Artificial Intelligence Solutions With Applications To Smart Environments, Fabrizio De Vita, Dario Bruneo, Sajal K. Das

Computer Science Faculty Research & Creative Works

Recent advances in Artificial Intelligence (AI) and the increasing availability of computational power have accelerated the diffusion of Intelligent Cyber-Physical Systems (ICPSs), enabling smart applications with reasoning capabilities. However, the limited resources of embedded and Edge devices significantly constrain the complexity of deep learning models that can be effectively deployed. Traditional approaches rely on cloud-based training and edge-only inference, a paradigm that becomes inadequate when low latency, privacy, security, and high customization are required. In this context, On-device AI is emerging as a new paradigm in which both training and inference are performed directly on the device, avoiding data transfer …

Real-Time Isolated Asl Recognition: Evaluating Spatial-Temporal Networks And Multimodal Llms, Raga Mouni Batchu

West Chester University Graduate Theses, Dissertations, and Final Projects

This thesis investigates the deployment of high-accuracy Isolated ASL Recognition (ISLR) in resource-constrained edge environments. We train a lightweight Spatio-Temporal Attention Network (SSTAN,∼2.7 M parameters,∼10 MB) on the WLASL-100 benchmark, achieving 75.25% Top-1 and 88.24% Top-5 accuracy with 139 ms CPU-only inference. A systematic comparison against frontier multimodal LLMs (Gemini 3 Flash, Gemini 3.1 Pro, Qwen 3 VL) shows SSTAN outperforms the best LLM baseline by∼1.85×in accuracy while being 22–230×faster and up to 40×cheaper annually. The LLMs’ core limitation is a lack of fine-grained temporal perception; they impose English-language semantic priors rather than learning the articulatory distinctions that define ASL …

A Unified Framework For Evaluating Training Efficiency In Deep (Bayesian) Neural Networks: Metrics, Overtraining, Stopping Criteria, And Grokking Computer Science, Eduardo Cueto Mendoza

Doctoral

Measuring training efficiency for artificial neural networks is an open research problem, current literature reports several attempts to define measures or create reporting frameworks. Current methods lack generality as they require measurements of the hardware or software thus, comparing efficiency between different systems can be difficult. Similarly, current metrics or frameworks generally do not propose the use of the metrics to directly improve training efficiency. This thesis presents three main contributions: (1) a novel framework that quantifies the training efficiency of a neural architecture on a learning task as the average ratio of model accuracy to total energy consumption during …

Evaluating Large Language Models For Line-Level Vulnerability Localization, Jian Zhang, Chong Wang, Anran Li, Weisong Sun, Cen Zhang, Wei Ma, Yang Liu

Research Collection School Of Computing and Information Systems

Recently, Automated Vulnerability Localization (AVL) has attracted growing attention, aiming to facilitate diagnosis by pinpointing the specific lines of code responsible for vulnerabilities. Large Language Models (LLMs) have shown potential in various domains, yet their effectiveness in line-level vulnerability localization remains underexplored. In this work, we present the first comprehensive empirical evaluation of LLMs for AVL. Our study examines 19 leading LLMs suitable for code analysis, including ChatGPT and multiple open-source models, spanning encoder-only, encoder-decoder, and decoder-only architectures, with model sizes from 60M to 70B parameters. We evaluate three paradigms including few-shot prompting, discriminative fine-tuning, and generative fine-tuning with and …

Modeling And Mitigating Atmospheric Degradation In Computer Vision With Application In Renewable Energy Prediction, Sumit Laha

Graduate Studies Theses and Dissertations 2026

Weather-induced variability poses significant challenges to the reliability and performance of modern computational systems, particularly those relying on visual perception and environmental prediction. This dissertation focuses on enhancing computer vision and machine learning based predictive models that operate under varying atmospheric conditions. Two representative weather-impacted applications are investigated: image dehazing and solar photovoltaic (PV) power output forecasting. Image dehazing focuses on the restoration of clear, unobstructed visuals from hazy or foggy images, a task that is vital for various applications. On the other hand, photovoltaic (PV) power forecasting aims to predict future solar energy generation based on historical sky images …

Robust Deep Learning One-Class Classification, Shahd Alnofaie

Graduate Studies Theses and Dissertations 2026

One-Class Classification (OCC) focuses on learning the characteristics of normal data and identifying observations that deviate from this learned pattern as anomalies. It is commonly used in applications such as medical diagnosis, cybersecurity, industrial monitoring, and fraud detection, where abnormal examples are often rare or unavailable during training. Classical approaches such as SVDD and LS-SVDD describe normal data using a hypersphere. While effective in some settings, these methods rely on shallow representations and can be sensitive to noise and contaminated observations. To address these limitations, this dissertation introduces a Deep LS-SVDD framework that combines hypersphere-based data description with deep neural …

Adaptive Deep Learning In Physical Layer Applications, Ali Owfi

All Dissertations

Traditionally, signal processing models in communication systems have been designed based on solid foundations in statistics and information theory, often assuming linearity and optimizing for simplified models. However, real-world communication systems exhibit numerous imperfections and non-linearities that traditional linear models struggle to capture accurately. Deep Learning (DL)-based approaches, unconstrained by rigid mathematical models, have shown promise in optimizing system performance by accommodating specific hardware configurations and dynamic channel conditions as an alternative to the traditional methods. Despite all the recent research efforts on DL-based methods for physical layer applications, DL models have still not been widely applied to physical layer …

Detection Of Phase-Binning And Interpolation Artifacts In 4-Dimensional Computed Tomography Imaging Using Deep Learning And Rule-Based Approaches, Jorge Cisneros, Nathan H. Feldt, Yevgeniy Vinogradskiy, Richard Castillo, Edward Castillo

Department of Radiation Oncology Faculty Papers

BACKGROUND: Four-dimensional computed tomography (4DCT) imaging is a crucial component to lung cancer radiotherapy planning and enables CT-ventilation-based functional avoidance planning to mitigate radiation toxicity. However, 4DCT scans are frequently impaired by acquisition artifacts that corrupt downstream analyses that depend on lung segmentation and deformable image registration, such as CT-ventilation and dose accumulation.

PURPOSE: This study develops 3D deep learning models to identify phase-binning artifacts at the voxel level and a heuristic, rule-based method to identify interpolation slices within 4DCT images.

METHODS: We introduce a generator that systematically inserts synthetic phase-binning and interpolation artifacts into any artifact-free breathing phase obtained …

Cssa-Fusion: Channel Selective And Spatial Alignment Infrared-Visible Image Fusion, Zhen Li, Zhi Zeng, Zhongrui Xiao, Ming Wen, Zhiyuan Zhang, Yibin Tian

Research Collection School Of Computing and Information Systems

Infrared-visible image fusion aims to integrate complementary information from two modalities to generate images with enriched semantic content. However, existing methods often neglect two critical aspects: the design of a local–global feature enhancement architecture and spatial alignment. To address these challenges, we propose Channel Selective and Spatial Alignment Fusion (CSSA-Fusion), a novel framework composed of two synergistic modules. The first is a selective channel and redundancy suppression module, which introduces a dual-branch selective channel attention mechanism to jointly capture local saliency and global channel importance for enhanced feature representation, and an informativeness–redundancy separation strategy to suppress redundant information while preserving …

Exploiting The In-Distribution Embedding Space With Deep Learning And Gaussian Discriminant Analysis For An Out-Of-Distribution Malware Attach Detection, Tosin Olusola Ige

Open Access Theses & Dissertations

State-of-the-art machine and deep learning models generally perform well on previously seen data, albeit with wrong close world assumption that all real-world data are from previously seen train and validation samples, hence there poor performance when exposed to data which deviates from previously seen training and validation set. This is clearly evident in the domain of cybersecurity where the world continues to experience several high profile malware attacks despite advancement in state-of-the-art research. The reason being that the constant evolvement of innovation in the development of tools and method deployed to carry out various attacks had given hackers and other …

Cnn Based Deep Learning Modeling With Explainability Analysis For Detecting Fraudulent Blockchain Transactions, Mohammad Hasan, Mohammad Shahriar Rahman, Mohammad Jabed Morshed Chowdhury, Iqbal H. Sarker

Research outputs 2022 to 2026

In the era of growing cryptocurrency adoption, Blockchain has emerged as a leading player in the digital payment landscape. However, this widespread popularity also brings forth various security challenges, including the need to safeguard against fraudulent activities. One of the paramount challenges in this regard is the detection of fraudulent transactions within the realm of Bitcoin data. This task significantly influences the trust and security of digital payments. Yet, it's a formidable challenge given the relatively low occurrence of fraudulent Bitcoin transactions. While deep learning techniques have demonstrated their prowess in fraud detection, there remains a scarcity of studies exploring …

Artificial Intelligence For Reliability: Predictive Health Maintenance And Geolocation In Gps-Denied Environments, Rafael Toche Pizano

Graduate Theses and Dissertations

In this dissertation, we explore the potential of machine learning and deep learning techniques to enhance the performance and robustness of applications across two major domains. By addressing the challenges within these fields, we demonstrate that we can leverage learning algorithms to obtain substantial improvements in accuracy and robustness. First, we tackle a problem in the field of predictive health maintenance. We propose a novel auto encoder and neural network based methodology to predict failure times in complex aviation systems to learn to distinguish between normal and abnormal operational behavior, and use this information to inform the neural network to …

Spatio-Temporal Gcn With Softmax Classifier For Skeleton-Based Human Action Recognition, Kabul Khudaybergenov, Avazjon Marakhimov, Zahriddin Muminov

Chemical Technology, Control and Management

Skeleton-based human action recognition is an important research area with many practical applications. Most existing methods rely on single representations of skeletal sequences, which cannot totally obtain all the complex features of human movements. This paper presents LFHAR (Latent Features for Human Action Recognition), a new framework that uses multiple spatio-temporal latent representations to improve the extraction of action features. Our method captures how skeletal poses change over time and combines motion information from both individual joints and connected body parts. The proposed approach applies graph-based processing to each skeleton frame in a sequence, then arranges the resulting graph features …

Zero Trust - The Future Of Cybersecurity, Danial Haider, Shougfta Mushtaq, Hasnat Ali, Mazliham Mohd Su’Ud

Journal of Informatics and Web Engineering

The recent Zero-Trust Architecture (ZTA) is progressively adopted to the develop network security by assuming no implicit trust within or outside an organization’s boundary. Though, ZTA faces substantial challenges in detecting sophisticated and developing cyber threats, particularly due to its trust on traditional security mechanisms that struggle to manage internal threats and sophisticated attack techniques. To report these shortcomings, the proposed study discovers the combination of advanced machine learning (ML) and deep learning (DL) performances to improve the anomaly detection proficiencies within ZTA environments. The study develops the CICIDS2017 dataset, which contains diverse and realistic network traffic patterns, to assess …

Deep Learning Approaches To Autocorrelation Function And Signal-To-Noise Ratio Estimation In Noisy Images, Kai Liang Lew, Kok Swee Sim, Shing Chiang Tan

Journal of Informatics and Web Engineering

Accurate estimation of signal-to-noise ratio (SNR) in Scanning Electron Microscopy (SEM) is crucial because it evaluates the image quality. SEM images faced a challenge whereby Gaussian noise commonly appears in the images. Thus, researchers have developed several methods to estimate the SNR value. With the introduction of deep learning, most of the limitations in the classical methods can be addressed. This paper proposes a novel deep learning, CNN-based Calibration Map Network (CalibNet) to estimate the SNR value from SEM images using a calibration map between classical SNR and autocorrelation function SNR. The architecture consists of convolutional layers, rectified linear unit …

Skin Cancer Image Classification Using Deep Learning With Data Segmentation Technique, Akhilesh Kumar Shrivas, Hema Vastrakar

Karbala International Journal of Modern Science

The human skin is an impressive organ and structural element often impacted by a diverse range of recognized and unknown diseases. Diagnosing disorders that affect the outermost layer of the body is the most uncertain and difficult component in the scientific field. Dermatological diseases are one of the most significant health concerns in the 21st century since their identification is challenging and costly, plagued with challenges and the subjectivity that comes with human interpretation. The main objective of this piece of research work is to develop a robust model for the classification of skin cancer diseases using deep convolution neural …

Bifusenet: A Multimodal Network For Estimating Blood Alcohol Concentration Via Bidirectional Hierarchical Fusion, Abdullah Tariq, Arooba Maqsood, Martin Masek, Syed Zulqarnain Gilani

Research outputs 2022 to 2026

Drunk driving remains a significant public safety challenge, demanding innovative alternatives to conventional methods such as field sobriety tests and breathalysers. Estimating a driver's level of intoxication through facial cues is particularly challenging due to the subtle and person-specific nature of alcohol-induced behaviours. In this paper, we present BiFuseNet, a 3D spatio-temporal multi-modal network designed to classify alcohol impairment levels into three categories: sober, moderate, and severe. Unlike prior approaches that rely on either uni-modal RGB video or hand-crafted facial features, our method exploits complementary physiological cues from RGB and infrared (IR) facial videos. We introduce a Bi-directional Hierarchical Fusion …

Detecting Polar Ring Galaxies Via Deep Learning, Fawad Kirmani, Anathavishnu S. Unnii, Varsha P. Kulkarni, Kyle Lackey, John R. Rose

Faculty Publications

Polar ring galaxies (PRGs) are peculiar galaxies that show a ring of stars, gas, and dust oriented roughly over the poles of the central ‘host’ galaxy (i.e. roughly orthogonal to the disc of the host galaxy). The formation models for these rings involve mergers or tidal interactions of the host galaxy with another galaxy. Although the identified PRGs look different from each other, they all have a ring that is not in the same plane as the disc of the host galaxy. Unlike in galaxies such as our Milky Way, where stars form in spiral arms, the rings exemplify an …

Application Of Machine Learning For Vascular System Analysis, Alireza Bagheri Rajeoni

Theses and Dissertations

The analysis of vascular structures is critical for diagnosing, monitoring, and treating vascular diseases such as aneurysms, stenosis, and vascular calcification. Traditional methods often rely on manual interpretation of imaging data, which is time-consuming, subjective, and not scalable. This work explores the application of advanced machine learning techniques to automate and enhance vascular system analysis. Our contributions include achieving state-of-the-art accuracy in vascular segmentation, developing a machine learning pipeline to automatically quantify vascular calcification in peripheral arterial disease, and designing a multi-stage machine learning system for abdominal aortic aneurysm analysis that identifies aneurysm boundaries and estimates aneurysm volume in a …

Ai-Enabled Multi-Layer Security Operations: A Combined Siem, Ids, And Threat Intelligence Model For Adaptive Cyber Defense, Mohamad Khayat

Dissertations

This dissertation presents a comprehensive framework for the evolution of Security Operation Centers (SOCs) through the integration of advanced artificial intelligence (AI), blockchain, and optimization techniques. Motivated by the increasing complexity of cyber threats and the limitations of traditional reactive SOC strategies, this work begins with a systematic literature review that identifies critical gaps in current SOC operations. Based on these insights, a reference architecture is proposed to guide the integration of intelligent components into SOC environments. To address the challenge of secure and trustworthy information sharing, a blockchain-based threat intelligence platform is developed, leveraging Byzantine Fault Tolerance and Zero-Knowledge …

Topoimages: Incorporating Local Topology Encoding Into Deep Learning Models For Medical Image Classification, Pengfei Gu, Hongxiao Wang, Yejia Zhang, Huimin Li, Chaoli Wang, Danny Z. Chen

Computer Science Faculty Publications

Topological structures in image data, such as connected components and loops, play a crucial role in understanding image content (e.g., biomedical objects). Despite remarkable successes of numerous image processing methods that rely on appearance information, these methods often lack sensitivity to topological structures when used in general deep learning (DL) frameworks. In this paper, we introduce a new general approach, called TopoImages (for Topology Images), which computes a new representation of input images by encoding local topology of patches. In TopoImages, we leverage persistent homology (PH) to encode geometric and topological features inherent in image patches. Our main objective is …

Domain Obedient Deep Learning, Soumadeep Saha

Doctoral Theses

Deep learning, a family of data-driven artificial intelligence techniques, has shown immense promise in a plethora of applications, and it has even outpaced experts in several domains. However, unlike symbolic approaches to learning, these methods fall short when it comes to abiding by and learning from pre-existing established principles. This is a significant deficit for deployment in critical applications such as robotics, medicine, industrial automation, etc. For a decision system to be considered for adoption in such fields, it must demonstrate the ability to adhere to specified constraints, an ability missing in deep learning-based approaches. Exploring this problem serves as …

A Deep Learning Framework With Explainable Ai For Atmospheric Blocking Detection And Interpretation, Devansh Khandelwal

Discovery Undergraduate Interdisciplinary Research Internship

Atmospheric blocking is a large-scale quasi-stationary phenomenon in mid-latitude circulation, characterized by persistent high-pressure systems that disrupt the typical west-to-east flow of the jet stream. These systems can cause extreme weather events—such as heatwaves, cold spells, or droughts—that persist for days or even weeks. This study proposes a deep learning framework to predict and interpret the occurrence of atmospheric blocking by integrating geophysical precursors such as geopotential height (Z500), stream function (SF200), and potential vorticity. These features, which are dynamically linked to blocking onset and persistence, serve as inputs to a Convolutional Neural Network model trained on the CESM Large …

Effective Transformer Networks For Undersampled Magnetic Resonance Image Reconstruction, Tahsin Rahman

Open Access Theses & Dissertations

The proliferation of data-driven tools for solving problems in every possible domain, coupled with rapid advances in computing technology, has led to an arms race of AI development and application research in industry and academia. One field of research that stands to gain immeasurably from this revolution is medical imaging. It is a critical part of modern diagnostics, and advancements in this area can directly benefit the average person by making healthcare more accessible, accurate, and affordable. Breakthroughs in mainstream image processing and computer vision have long fueled development in medical imaging, and it is now common to see cutting …

Automated Chick Sexing Using Computer Vision, Marta Veganzones Rodriguez

Graduate Theses and Dissertations

This thesis presents two complementary approaches to chick sexing, a critical task in poultry production that demands accurate and early gender identification. By investigating both facial and vent-based modalities, we present two complementary methods that aim to improve the efficiency, scalability, and ethical standards of gender classification in day-old chicks through the use of computer vision and deep learning. The first approach draws inspiration from human facial gender recognition to introduce facial chick sexing, a minimally invasive technique that eliminates the need for expert knowledge. This system encompasses a complete pipeline that includes image acquisition, facial detection and alignment, keypoint …

Revolutionizing Alzheimer’S Diagnosis: A Hybrid Deep Learning Approach For Enhanced Mri Analysis, Hassan A. Ahmed, Syed H. Ahmed, Kyle Nash

Business Faculty Publications

Alzheimer’s Disease (AD) is a neurodegenerative disorder that primarily affects the elderly, causing cognitive decline and memory loss. Traditional diagnostic methods, such as neuropsychological tests and cerebrospinal fluid analysis, are invasive and time-consuming. Neuroimaging techniques like MRI and PET provide valuable insights but require manual analysis by specialists. This study proposes a hybrid model combining EfficientNetB0, a deep learning architecture, with Convolutional Neural Networks (CNN) to automate AD detection in MRI scans. The model uses data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset, which includes over 200 MRI scans and clinical information. Our results show that the hybrid model …

Input Structure Based Optimization For Privacy Preserving Ai Systems, Feng Yizhou

Electrical & Computer Engineering Theses & Dissertations

As Artificial Intelligence (AI) systems become increasingly integrated into critical domains, ensuring privacy-preserving model design and system deployment has become a pressing priority. Safeguarding both sensitive user data and proprietary model parameters is critical throughout the AI model and system, from data acquisition and pre-processing to model inference and deployment. However, existing privacy-preserving frameworks face several limitations, including fragmented data ownership, incomplete protection across system stages, substantial computational overhead, and poor scalability to modern architectures such as large language models. This dissertation explores a unifying optimization strategy centered on input structure design to address these challenges. The core idea is …