Computer Sciences Commons^™

Advancing Household Robotics: Deep Interactive Reinforcement Learning For Efficient Training And Enhanced Performance, Arpita Soni, Sujatha Alla, Suresh Dodda, Hemanth Volikatla

Engineering Management & Systems Engineering Faculty Publications

The market for domestic robots—made to perform household chore, is growing as these robots relieve people of everyday responsibilities. Domestic robots are generally welcomed for their role in easing human labour, in contrast to industrial robots, which are frequently criticised for displacing human workers. But before these robots can carry out domestic chores, they need to become proficient in a number of minor activities, such as recognizing their surroundings, making decisions, and picking up on human behaviours. Reinforcement learning, or RL, has emerged as a key robotics technology that enables robots to interact with their environment and learn how to …

Improving Neuropathological Analysis With Aβgan: Addressing Morphology Imbalance For Efficient Alzheimer's Disease Diagnosis, Sujatha Alla, Prasanthi Chidipudi, Nagesh Bheesetty, Vedvikash Reddy Velur, Joshit Mohanty, Puneeth Bheesetty, Marisha Jmukhadze, Narendra Lakshmana Gowda, Sai Gireesh Komaragiri

Engineering Management & Systems Engineering Faculty Publications

Histopathologists are experiencing a digital revolution in their field thanks to the digitization of Whole Slide Images (WSIs), which are microscope slides of tissue that can measure gigapixels in size. With so much high resolution data at their disposal, computer vision techniques can now be used to automate laboratory processes, create visual standards, and increase analysis throughput, all of which reduce the workload of pathologists [1]. The "gold" standard in neuropathology, particularly for Alzheimer's Disease- is pathological diagnosis made by looking at White Matter Inclusions (WSIs) in brain tissue. Semi-quantitative scoring in accordance with the standards established by the Consortium …

Context-Free Grammar Framework For Automatic Shooting Game Enemy Pattern Generation, Nitit Kaweeratanakit

Chulalongkorn University Theses and Dissertations (Chula ETD)

This research proposes a framework for generating enemy patterns for SHMUPs game. It is directly based on a grammar derived from the enemy behavior of existing commercial SHMUPs, and implemented using a new description language called "Enemy Pattern Description Language" (EPDL). EPDL contains all information required to construct the enemy, with no requirement of external data content. The language is human-readable and can be connected to any game engine of choice using an EPDL interpreter. The interpreter itself consists of lexer and recursive descent parser. The results shown in this research is implemented in. "rdnh", a private fork of Touhou …

การพยากรณ์การจ่ายยาของโรงพยาบาลโดยการเรียนรู้ของเครื่องและวิธีการวิเคราะห์เชิงสถิติ, วริศ ปุณณะหิตานนท์

Chulalongkorn University Theses and Dissertations (Chula ETD)

ในปัจจุบันการพยากรณ์การจ่ายยาของโรงพยาบาลถือเป็นหัวใจสำคัญต่อการจัดการคลังยาและการสั่งซื้อยาเป็นอย่างมาก เนื่องจากการพยากรณ์ที่น้อยเกินไปทำให้ยาไม่เพียงพอส่งผลให้เกิดความล่าช้าภายในโรงพยาบาล ในขณะที่การพยากรณ์ที่มากเกินไปทำให้เปลืองพื้นที่ใช้สอยและอาจทำให้ยาเสื่อมสภาพหรือหมดอายุซึ่งส่งผลให้โรงพยาบาลสูญเสียรายได้ การมีแบบจำลองที่สามารถพยากรณ์ปริมาณการจ่ายยาให้ใกล้เคียงกับค่าจริงจะสามารถลดปัญหาการขาดแคลนยาในแต่ละห้องจ่ายยาหรือการที่ห้องจ่ายยามีการกักตุนตัวยาเกินความจำเป็น จากปัญหาที่กล่าวมาข้างต้น โครงงานมหาบัณฑิตนี้จึงถูกจัดทำขึ้นเพื่อนำเสนอแบบจำลองที่จะมาแทนค่าเฉลี่ยเคลื่อนที่แบบทั่วไปซึ่งจะช่วยให้โรงพยาบาลสามารถพยากรณ์ปริมาณการจ่ายยาแต่ละวันได้แม่นยำมากขึ้น โดยจะนำเทคนิคสำหรับพยากรณ์ข้อมูลที่อยู่ในรูปแบบของอนุกรมเวลามาประยุกต์ใช้กับข้อมูลการจ่ายยาย้อนหลังและข้อมูลการนัดหมายแพทย์ย้อนหลัง หลังจากนั้นจะนำผลลัพธ์ที่ได้มาคำนวณค่าเคลาดเคลื่อนด้วยค่าเฉลี่ยของเปอร์เซ็นต์ความคลาดเคลื่อนสัมบูรณ์และค่าเฉลี่ยสมมาตรของเปอร์เซ็นต์ความคลาดเคลื่อนสัมบูรณ์และนำผลที่ได้มาใช้ในการเลือกว่าแบบจำลองไหนให้ค่าคลาดเคลื่อนต่ำที่สุด ผลการทดลองพบว่าแบบจำลองซัพพอร์ตเวกเตอร์รีเกรสชันให้ค่าความคลาดเคลื่อนที่ต่ำกว่าค่าเฉลี่ยเคลื่อนที่แบบทั่วไป แบบจำลองที่ผู้จัดทำโครงงานนำเสนอสามารถนำไปประยุกต์ใช้กับการพยากรณ์การจ่ายยาเพื่อให้แต่ละห้องจ่ายยามียาสำหรับให้บริการในปริมาณที่เพียงพอต่อความต้องการ

Uncertainty Quantification In Large Language Models Through Convex Hull Analysis, Ferhat Ozgur Catak, Murat Kuzlu

Engineering Technology Faculty Publications

Uncertainty quantification approaches have been more critical in large language models (LLMs), particularly high-risk applications requiring reliable outputs. However, traditional methods for uncertainty quantification, such as probabilistic models and ensemble techniques, face challenges when applied to the complex and high-dimensional nature of LLM-generated outputs. This study proposes a novel geometric approach to uncertainty quantification using convex hull analysis. The proposed method leverages the spatial properties of response embeddings to measure the dispersion and variability of model outputs. The prompts are categorized into three types, i.e., ’easy’, ’moderate’, and ’confusing’, to generate multiple responses using different LLMs at varying temperature settings. …

Federated Learning: Overview, Strategies, Applications, Tools And Future Directions, Betul Yurdem, Murat Kuzlu, Mehmet Kemal Gullu, Maliha Tabassum

Engineering Technology Faculty Publications

Federated learning (FL) is a distributed machine learning process, which allows multiple nodes to work together to train a shared model without exchanging raw data. It offers several key advantages, such as data privacy, security, efficiency, and scalability, by keeping data local and only exchanging model updates through the communication network. This review paper provides a comprehensive overview of federated learning, including its principles, strategies, applications, and tools along with opportunities, challenges, and future research directions. The findings of this paper emphasize that federated learning strategies can significantly help overcome privacy and confidentiality concerns, particularly for high-risk applications.

Towards Algorithmic Justice: Human Centered Approaches To Artificial Intelligence Design To Support Fairness And Mitigate Bias In The Financial Services Sector, Jihyun Kim

CMC Senior Theses

Artificial Intelligence (AI) has positively transformed the Financial services sector but also introduced AI biases against protected groups, amplifying existing prejudices against marginalized communities. The financial decisions made by biased algorithms could cause life-changing ramifications in applications such as lending and credit scoring. Human Centered AI (HCAI) is an emerging concept where AI systems seek to augment, not replace human abilities while preserving human control to ensure transparency, equity and privacy. The evolving field of HCAI shares a common ground with and can be enhanced by the Human Centered Design principles in that they both put humans, the user, at …

Ai-Based Defect Detection In Aerospace Ultrasonic Signals, Rami Issac Lake

Graduate Research Theses & Dissertations

Ensuring the safety and integrity of materials and structures throughout the manufacturing cycle is a critical concern across various industries, including aerospace, automotive, oiland gas, and civil engineering. Non-Destructive Inspection (NDI) techniques allow for the examination of materials without causing damage or alteration, enabling the early detection of potential issues before materials are utilized in the field. The inspection of fuselage composites presents a particular challenge due to their complex structures, diverse materials, and differences in thickness, making defect detection a challenging yet crucial task. Moreover, defects of various types and causes can emerge across all depths of the material …

Simulation Of A Pick And Place System For Electronic Cards Using A Yumi Cobot, Derrick Sze, Rosula Sj Reyes, Patricia Angela R. Abu

Electronics, Computer, and Communications Engineering Faculty Publications

Collaborative Robots are one of the main drivers of Industry 4.0, which started as a vision focusing on industrial production. It addresses several challenges in the current manufacturing industry such as performing repetitive work and requiring highly skilled workers. The goal of the research is to be able to simulate a pick and place environment with electronic cards using a YuMi cobot and mobile platforms in Coppeliasim. The mobile robot is responsible for transporting the electronic cards to the target location through path planning implemented using the OMPL plug-in. After arriving at the target location, YuMi will then perform the …

Machine Learning Assisted Optimization For Calculation And Automated Tuning Of Antennas, Lauren Linkous

Theses and Dissertations

The Antenna Calculation and Autotuning Tool (AntennaCAT) software suite represents a significant advancement in the field of antenna design by automating the entire design, CAD, simulation, and optimization process compatible with several EM simulation software suites. It is the first comprehensive implementation of machine learning in this context. In particular, this work includes the capability to create and export structured datasets from the aforementioned EM software for iterative improvement and includes an expandable selection of optimizers.

Evaluating Computational Reproducibility Of Jupyter Notebooks Using Machine Learning And Natural Language Processing, A S M Shahadat Hossain

Graduate Research Theses & Dissertations

In recent years, computational reproducibility, which refers to achieving consistent results upon rerunning an experiment, has become one of the major concerns of various research communities. Jupyter Notebook, as a web-based computational notebook application, offers useful features for running and publishing computational experiments through interactive environments. However, rerunning notebooks does not always reproduce the experimental results. This thesis aims to develop novel methods to evaluate reproducibility by comparing different types of outputs between original and rerun notebooks. It also explores the idea of using machine learning models to predict reproducibility of notebooks automatically without the need of rerunning them. Through …

Bidirectional In-Situ Analysis And Visualization Of Fluid Structure Interactions With Applications In Blood Flow, Nazariy Tishchenko

Graduate Research Theses & Dissertations

Interactive blood flow simulation has applications in the understanding of complex fluids,training medical professionals, diagnostics, and surgical procedure planning. Bidirectional in situ visualization and analysis are much needed to streamline the analysis of large-scale patient-specific blood flow simulations. The in situ component ensures the live visualization of the simulation results while the simulation is running, and the bidirectional component allows for real-time control and configuration of the simulation parameters (i.e., setting and retrieving the simulation parameters and data based on need). In this project, a highly accurate multiphysics simulation was instrumented to work with an existing bidirectional in situ framework. …

Flexible Fitting Of Alphafold2-Predicted Models To Cryo-Em Density Maps Using Elastic Network Models: A Methodological Affirmation, Maytha Alshammari, Jing He, Willy Wriggers

Computer Science Faculty Publications

Motivation: This study investigates the flexible refinement of AlphaFold2 models against corresponding cryo-electron microscopy (cryo-EM) maps using normal modes derived from elastic network models (ENMs) as basis functions for displacement. AlphaFold2 generally predicts highly accurate structures, but 18 of the 137 models of isolated chains exhibit a TM-score below 0.80. We achieved a significant improvement in four of these deviating structures and used them to systematically optimize the parameters of the ENM motion model.

Results: We successfully refined four AlphaFold2 models with notable discrepancies: lipid-preserved respiratory supercomplex (TM-score increased from 0.52 to 0.69), flagellar L-ring protein (TM-score increased from 0.53 …

A Survey On Few-Shot Class-Incremental Learning, Songsong Tian, Lusi Li, Weijun Li, Hang Ran, Xin Ning, Prayag Tiwari

Computer Science Faculty Publications

Large deep learning models are impressive, but they struggle when real-time data is not available. Few-shot class-incremental learning (FSCIL) poses a significant challenge for deep neural networks to learn new tasks from just a few labeled samples without forgetting the previously learned ones. This setup can easily leads to catastrophic forgetting and overfitting problems, severely affecting model performance. Studying FSCIL helps overcome deep learning model limitations on data volume and acquisition time, while improving practicality and adaptability of machine learning models. This paper provides a comprehensive survey on FSCIL. Unlike previous surveys, we aim to synthesize few-shot learning and incremental …

Speculative Anisotropic Mesh Adaptation On Shared Memory For Cfd Applications, Christos Tsolakis, Nikos Chrisochoides

Computer Science Faculty Publications

Efficient and robust anisotropic mesh adaptation is crucial for Computational Fluid Dynamics (CFD) simulations. The CFD Vision 2030 Study highlights the pressing need for this technology, particularly for simulations targeting supercomputers. This work applies a fine-grained speculative approach to anisotropic mesh operations. Our implementation exhibits more than 90% parallel efficiency on a multi-core node. Additionally, we evaluate our method within an adaptive pipeline for a spectrum of publicly available test-cases that includes both analytically derived and error-based fields. For all test-cases, our results are in accordance with published results in the literature. Support for CAD-based data is introduced, and its …

Autonomous Strike Uavs In Support Of Homeland Security Missions: Challenges And Preliminary Solutions, Meshari Aljohani, Ravi Mukkamala, Stephan Olariu

Computer Science Faculty Publications

Unmanned Aerial Vehicles (UAVs) are becoming crucial tools in modern homeland security applications, primarily because of their cost-effectiveness, risk reduction, and ability to perform a wider range of activities. This study focuses on the use of autonomous UAVs to conduct, as part of homeland security applications, strike missions against high-value terrorist targets. Owing to developments in ledger technology, smart contracts, and machine learning, activities formerly carried out by professionals or remotely flown UAVs are now feasible. Our study provides the first in-depth analysis of the challenges and preliminary solutions for the successful implementation of an autonomous UAV mission. Specifically, we …

Mosaic: A Prune-And-Assemble Approach For Efficient Model Pruning In Privacy-Preserving Deep Learning, Yifei Cai, Qiao Zhang, Rui Ning, Chunsheng Xin, Hongyi Wu

Computer Science Faculty Publications

To enable common users to capitalize on the power of deep learning, Machine Learning as a Service (MLaaS) has been proposed in the literature, which opens powerful deep learning models of service providers to the public. To protect the data privacy of end users, as well as the model privacy of the server, several state-of-the-art privacy-preserving MLaaS frameworks have also been proposed. Nevertheless, despite the exquisite design of these frameworks to enhance computation efficiency, the computational cost remains expensive for practical applications. To improve the computation efficiency of deep learning (DL) models, model pruning has been adopted as a strategic …

Enhancing Research Productivity: Seamless Integration Of Personal Devices And Hpc Resources With The Cybershuttle Notebook Gateway, Yasith Jayawardana, Dimuthu Wannipurage, Eroma Abeysinghe, Suresh Marru

Computer Science Faculty Publications

Scientists often utilize personal laptops and workstations for initial research stages and turn to high-performance computing (HPC) supercomputers for compute-intensive tasks. However, seamless transitions between these environments are vital for enhancing productivity and accelerating research progress. Our paper presents the Cybershuttle Notebook Gateway, an open-source framework crafted to streamline this transition, optimize resource utilization, and reduce time-to-science for researchers. Leveraging JupyterLab, the framework extends kernel mechanics for seamless provisioning and connection to remote HPC cluster kernels. We delve into its architecture, which separates user authentication, kernel provisioning, and remote file system access. Additionally, we highlight practical capabilities like analyzing network …

Enhanced Skin Cancer Diagnosis Through Grid Search Algorithm-Optimized Deep Learning Models For Skin Lesion Analysis, Rudresh Pillai, Neha Sharma, Sheifali Gupta, Deepali Gupta, Sapna Juneja, Saurav Malik, Hong Qin, Mohammed S. Alqahtani, Amel Ksibi

Computer Science Faculty Publications

Skin cancer is a widespread and perilous disease that necessitates prompt and precise detection for successful treatment. This research introduces a thorough method for identifying skin lesions by utilizing sophisticated deep learning (DL) techniques. The study utilizes three convolutional neural networks (CNNs)-CNN1, CNN2, and CNN3-each assigned to a distinct categorization job. Task 1 involves binary classification to determine whether skin lesions are present or absent. Task 2 involves distinguishing between benign and malignant lesions. Task 3 involves multiclass classification of skin lesion images to identify the precise type of skin lesion from a set of seven categories. The most optimal …

Disentangling Cyclic Causality: An Instance-Based Framework For Causal Discovery, Chase A. Yakaboski

Dartmouth College Ph.D Dissertations

Correlation does not imply causation" is one of the fundamental principles taught in science, emphasizing that associations between variables do not necessarily indicate causality. Yet, over the past three decades, extensive research has begun to challenge this perspective by developing sophisticated methods to differentiate causal from correlative relationships. This research suggests that correlations often involve a blend of confounded and causal interactions, which, given certain assumptions, can be disentangled to uncover actionable insights and deepen our understanding of physical, biological, and societal systems.

Accurately discovering causal relationships from data amidst cyclic dynamics remains a challenging open problem in causality research. …

A Memory Efficient Deep Recurrent Q-Learning Approach For Autonomous Wildfire Surveillance, Jeremy A. Cantor

UNF Graduate Theses and Dissertations

Previous literature demonstrates that autonomous UAVs (unmanned aerial vehicles) have the po- tential to be utilized for wildfire surveillance. This advanced technology empowers firefighters by providing them with critical information, thereby facilitating more informed decision-making processes. This thesis applies deep Q-learning techniques to the problem of control policy design under the objective that the UAVs collectively identify the maximum number of locations that are under fire, assuming the UAVs can share their observations. The prohibitively large state space underlying the control policy motivates a neural network approximation, but prior work used only convolutional layers to extract spatial fire information from …

An Analysis Of Precision: Occlusion And Perspective Geometry’S Role In 6d Pose Estimation, Jeffrey Choate, Derek Worth, Scott Nykl, Clark N. Taylor, Brett J. Borghetti, Christine M. Schubert Kabban

Faculty Publications

Achieving precise 6 degrees of freedom (6D) pose estimation of rigid objects from color images is a critical challenge with wide-ranging applications in robotics and close-contact aircraft operations. This study investigates key techniques in the application of YOLOv5 object detection convolutional neural network (CNN) for 6D pose localization of aircraft using only color imagery. Traditional object detection labeling methods suffer from inaccuracies due to perspective geometry and being limited to visible key points. This research demonstrates that with precise labeling, a CNN can predict object features with near-pixel accuracy, effectively learning the distinct appearance of the object due to perspective …

Transformer-Enabled Deep Reinforcement Learning For Coverage Path Planning, Daniel B. Tiu

UNF Graduate Theses and Dissertations

Coverage path planning (CPP) is the problem of covering all points in an environment and is a well-researched topic in robotics due to its sheer practical relevance. This paper investigates such an offline CPP problem where the primary objective is to minimize the path length to achieve complete coverage. Furthermore, the literature suggests that taking turns leads to a higher energy use than going straight. To this end, we design a novel objective function that aims to minimize the number of turns as well. We have proposed a deep reinforcement learning (DRL)-based framework that uses a Transformer model. Unlike state-of-the-art …

Adversarial Transferability And Generalization In Robust Deep Learning, Tao Wu

Doctoral Dissertations

Despite its remarkable achievements across a multitude of benchmark tasks, deep learning (DL) models exhibit significant fragility to adversarial examples, i.e., subtle modifications applied to inputs during testing yet effective in misleading DL models. These meticulously crafted perturbations possess the remarkable property of transferability: an adversarial example that effectively fools one model often retains its effectiveness against another model, even if the two models were trained independently. This research delves into the characteristics influencing the transferability of adversarial examples from three distinct and complementary perspectives: data, model, and optimization. Firstly, from the data perspective, we propose a new method of …

Ethical Decision-Making In Older Drivers During Critical Driving Situations: An Online Experiment, Amandeep Singh, Sarah Yahoodik, Yovela Murzello, Samuel Petkac, Yusuke Yamani, Siby Samuel

Psychology Faculty Publications

The present study examined the impact of aging on ethical decision-making in simulated critical driving scenarios. 204 participants from North America, grouped into two age groups (18–30 years and 65 years and above), were asked to decide whether their simulated automated vehicle should stay in or change from the current lane in scenarios mimicking the Trolley Problem. Each participant viewed a video clip rendered by the driving simulator at Old Dominion University and pressed the space-bar if they decided to intervene in the control of the simulated automated vehicle in an online experiment. Bayesian hierarchical models were used to analyze …

An Enhanced Deep Autoencoder For Flight Delay Prediction, Desmond B. Bisandu, Dan Andrei Soviani-Sitoiu, Irene Moulitsas

Journal of Aviation/Aerospace Education & Research

Accurate and timely flight delay prediction cannot be overemphasized because of the ever-increasing demand for air travel and its importance in deploying intelligent transportation systems. Nonetheless, there has not been a universal solution to the problem, as more intelligent flight decision systems are required for the aviation industry's future growth. Existing flight delay classification and prediction approaches are mainly shallow traffic models and do not satisfy many applications in the real world. Our motivation to rethink the deep architecture model for predicting flight delays emanates from the problem. In this research, we proposed a technique that modified stacked autoencoder architecture …

Sub-Band Backdoor Attack In Remote Sensing Imagery, Kazi Aminul Islam, Hongyi Wu, Chunsheng Xin, Rui Ning, Liuwan Zhu, Jiang Li

Electrical & Computer Engineering Faculty Publications

Remote sensing datasets usually have a wide range of spatial and spectral resolutions. They provide unique advantages in surveillance systems, and many government organizations use remote sensing multispectral imagery to monitor security-critical infrastructures or targets. Artificial Intelligence (AI) has advanced rapidly in recent years and has been widely applied to remote image analysis, achieving state-of-the-art (SOTA) performance. However, AI models are vulnerable and can be easily deceived or poisoned. A malicious user may poison an AI model by creating a stealthy backdoor. A backdoored AI model performs well on clean data but behaves abnormally when a planted trigger appears in …

Domain Adaptive Federated Learning For Multi-Institution Molecular Mutation Prediction And Bias Identification, W. Farzana, M. A. Witherow, I. Longoria, M. S. Sadique, A. Temtam, K. M. Iftekharuddin

Electrical & Computer Engineering Faculty Publications

Deep learning models have shown potential in medical image analysis tasks. However, training a generalized deep learning model requires huge amounts of patient data that is usually gathered from multiple institutions which may raise privacy concerns. Federated learning (FL) provides an alternative to sharing data across institutions. Nonetheless, FL is susceptible to a few challenges including inversion attacks on model weights, heterogenous data distributions, and bias. This study addresses heterogeneity and bias issues for multi-institution patient data by proposing domain adaptive FL modeling using several radiomics (volume, fractal, texture) features for O6-methylguanine-DNA methyltransferase (MGMT) classification across multiple institutions. The proposed …

Using Feature Selection Enhancement To Evaluate Attack Detection In The Internet Of Things Environment, Khawlah Harahsheh, Rami Al-Naimat, Chung-Hao Chen

Electrical & Computer Engineering Faculty Publications

The rapid evolution of technology has given rise to a connected world where billions of devices interact seamlessly, forming what is known as the Internet of Things (IoT). While the IoT offers incredible convenience and efficiency, it presents a significant challenge to cybersecurity and is characterized by various power, capacity, and computational process limitations. Machine learning techniques, particularly those encompassing supervised classification techniques, offer a systematic approach to training models using labeled datasets. These techniques enable intrusion detection systems (IDSs) to discern patterns indicative of potential attacks amidst the vast amounts of IoT data. Our investigation delves into various aspects …

Adversarial Training Based Domain Adaptation Of Skin Cancer Images, Syed Qasim Gilani, Muhammad Umair, Maryam Naqvi, Oge Marques, Hee-Cheol Kim

Electrical & Computer Engineering Faculty Publications

Skin lesion datasets used in the research are highly imbalanced; Generative Adversarial Networks can generate synthetic skin lesion images to solve the class imbalance problem, but it can result in bias and domain shift. Domain shifts in skin lesion datasets can also occur if different instruments or imaging resolutions are used to capture skin lesion images. The deep learning models may not perform well in the presence of bias and domain shift in skin lesion datasets. This work presents a domain adaptation algorithm-based methodology for mitigating the effects of domain shift and bias in skin lesion datasets. Six experiments were …