Other Computer Engineering Commons^™

Design Of Multi-Objective Optimization Algorithms For Vlsi Floor Planning, Srinivasan B

Theses and Dissertations

VLSI floorplanning is a key design step that determines the optimal placement of circuit modules to minimize chip area, wire length, and heat generation. Existing swarm intelligence–based metaheuristics improve area and wire length but often ignore thermal effects.

To address this, the Multi-Objective Firefly Optimization–based Floorplanning (MOFO-FP) technique is introduced, using a Heat-Aware Firefly Optimization (HAFO) algorithm that minimizes heat, space, and wire length under fixed outline constraints. Each firefly represents a floorplan, with brightness indicating solution quality; dimmer fireflies move toward brighter ones to find optimal placements.

A second method, the Hybridized Multicriteria Ant Colony and …

Toward Intuitive 3d Interactions In Virtual Reality: A Deep Learning- Based Dual-Hand Gesture Recognition Approach, Trudi Di Qi, Franceli L. Cibrian, Meghna Raswan, Tyler Kay, Hector M. Camarillo-Abad, Yuxin Wen

Engineering Faculty Articles and Research

Dual-hand gesture recognition is crucial for intuitive 3D interactions in virtual reality (VR), allowing the user to interact with virtual objects naturally through gestures using both handheld controllers. While deep learning and sensor-based technology have proven effective in recognizing single-hand gestures for 3D interactions, research on dual-hand gesture recognition for VR interactions is still underexplored. In this work, we introduce CWT-CNN-TCN, a novel deep learning model that combines a 2D Convolution Neural Network (CNN) with Continuous Wavelet Transformation (CWT) and a Temporal Convolution Network (TCN). This model can simultaneously extract features from the time-frequency domain and capture long-term dependencies using …

Beyond The Horizon: Exploring Anomaly Detection Potentials With Federated Learning And Hybrid Transformers In Spacecraft Telemetry, Juan Rodriguez

Computer Science and Engineering Theses and Dissertations

Telemetry sensors play a crucial role in spacecraft operations, providing essential data on efficiency, sustainability, and safety. However, identifying irregularities in telemetry data can be a time-consuming process that risks the success of missions. With the rise of CubeSats and smallsats, telemetry data has become more abundant, but concerns about privacy and scalability have resulted in untapped data potential. To address these issues, we propose a new approach to anomaly detection that utilizes machine learning models at data sources. These models solely transmit weights to a centralized server for aggregation, resulting in improved dataset performance with a single global model. …

Development Of Hybrid Multi Criteria Decision Making Techniques For Efficient Cloud Service Selection, Obulaporam Gireesha

Theses and Dissertations

During the past few decades, cloud computing became a primary driver for the next generation of digital technology due to the increase in organizational performance and profitability based on a ‘pay-as-you-use’ fashion at anytime and anywhere across the globe. Cloud computing enables various enterprises to access pooled resources (like storage, network bandwidth, software applications, processing power, etc.) over the Internet with minimal Information Technology (IT) infrastructure and capital expenditure.

Indeed, the enormous popularity of cloud computing in both academia & industry over the decade has resulted in a wide range of similar cloud services offered by numerous service providers. Even …

Neuro-Symbolic Commonsense Reasoning With Resistance To Data Poisoning: A First-Order Logic And Sub-Symbolic Embeddings Framework, Bryce Shurts, King-Ip Lin

Computer Science and Engineering Theses and Dissertations

Commonsense reasoning has long presented a hurdle between conversational agents and their ability to naturally engage with humans in conversation, as the infinitely dimensional nature of a dialogue’s topics presents a significant reasoning challenge in the study of Natural Language Understanding (NLU). Such a system must conceivably be able to act as a generalizable system for evaluating and reasoning about commonsense statements, problems, and queries: in this way, the agent can attempt to quantify the reasonability of a given input. We attempt to address this through the integration of an explainable neuro-symbolic system that leverages Logical Tensor Networks (LTNs) and …

Generalized Model To Enable Zero-Shot Imitation Learning For Versatile Robots, Yongshuai Wu

Master's Theses

The rapid advancement in Deep Learning (DL), especially in Reinforcement Learning (RL) and Imitation Learning (IL), has positioned it as a promising approach for a multitude of autonomous robotic systems. However, the current methodologies are predominantly constrained to singular setups, necessitating substantial data and extensive training periods. Moreover, these methods have exhibited suboptimal performance in tasks requiring long-horizontal maneuvers, such as Radio Frequency Identification (RFID) inventory, where a robot requires thousands of steps to complete.

In this thesis, we address the aforementioned challenges by presenting the Cross-modal Reasoning Model (CMRM), a novel zero-shot Imitation Learning policy, to tackle long-horizontal robotic …

Practical Static Binary Instrumentation Attacks Against Binary Stylometry, Justin Lee Carpenter

Masters Theses, 2020-current

Binary stylometry aims to find the features in a binary computer program and use them to identify the developers of the corresponding source code. Despite the noises in the code compilation process from the compiler, assembler, linker, and library functions, two existing studies based on machine learning for binary stylometry have reported high success rates (Alrabaee, Shirani, Wang, Debbabi, and Hanna 2018; Caliskan, Yamaguchi, Dauber, Harang, Rieck, Greenstadt, and Narayanan 2018). In this thesis, we first observe that both existing studies are based on a largely benign security model and assume that the binaries used in testing and prediction are …

Brain Computer Interface-Based Drone Control Using Gyroscopic Data From Head Movements, Ikaia Cacha Melton

Honors College Theses

This research explores the potential of using gyroscopic data from a person’s head movement to control a DJI Tello quadcopter via a Brain-Computer Interface (BCI). In this study, over 100 gyroscopic recordings capturing the X, Y and Z columns (formally known as GyroX, GyroY, GyroZ) between 4 volunteers with the Emotiv Epoc X headset were collected. The Emotiv Epoc X data captured (left, right, still, and forward) head movements of each participant associated with the DJI Tello quadcopter navigation. The data underwent thorough processing and analysis, revealing distinctive patterns in charts using Microsoft Excel. A Python condition algorithm was then …

Improving Ethics Surrounding Collegiate-Level Hacking Education: Recommended Implementation Plan & Affiliation With Peer-Led Initiatives, Shannon Morgan, Dr. Sanjay Goel

Military Cyber Affairs

Cybersecurity has become a pertinent concern, as novel technological innovations create opportunities for threat actors to exfiltrate sensitive data. To meet the demand for professionals in the workforce, universities have ramped up their academic offerings to provide a broad range of cyber-related programs (e.g., cybersecurity, informatics, information technology, digital forensics, computer science, & engineering). As the tactics, techniques, and procedures (TTPs) of hackers evolve, the knowledge and skillset required to be an effective cybersecurity professional have escalated accordingly. Therefore, it is critical to train cyber students both technically and theoretically to actively combat cyber criminals and protect the confidentiality, integrity, …

Using Digital Twins To Protect Biomanufacturing From Cyberattacks, Brenden Fraser-Hevlin, Alec W. Schuler, B. Arda Gozen, Bernard J. Van Wie

Military Cyber Affairs

Understanding of the intersection of cyber vulnerabilities and bioprocess regulation is critical with the rise of artificial intelligence and machine learning in manufacturing. We detail a case study in which we model cyberattacks on network-mediated signals from a novel bioreactor, where it is important to control medium feed rates to maintain cell proliferation. We use a digital twin counterpart reactor to compare glucose and oxygen sensor signals from the bioreactor to predictions from a kinetic growth model, allowing discernment of faulty sensors from hacked signals. Our results demonstrate a successful biomanufacturing cyberattack detection system based on fundamental process control principles.

Characterizing Advanced Persistent Threats Through The Lens Of Cyber Attack Flows, Logan Zeien, Caleb Chang, Ltc Ekzhin Ear, Dr. Shouhuai Xu

Military Cyber Affairs

Effective cyber defense must build upon a deep understanding of real-world cyberattacks to guide the design and deployment of appropriate defensive measures against current and future attacks. In this abridged paper (of which the full paper is available online), we present important concepts for understanding Advanced Persistent Threats (APTs), our methodology to characterize APTs through the lens of attack flows, and a detailed case study of APT28 that demonstrates our method’s viability to draw useful insights. This paper makes three technical contributions. First, we propose a novel method of constructing attack flows to describe APTs. This abstraction allows technical audiences, …

Machine Learning Security For Tactical Operations, Dr. Denaria Fields, Shakiya A. Friend, Andrew Hermansen, Dr. Tugba Erpek, Dr. Yalin E. Sagduyu

Military Cyber Affairs

Deep learning finds rich applications in the tactical domain by learning from diverse data sources and performing difficult tasks to support mission-critical applications. However, deep learning models are susceptible to various attacks and exploits. In this paper, we first discuss application areas of deep learning in the tactical domain. Next, we present adversarial machine learning as an emerging attack vector and discuss the impact of adversarial attacks on the deep learning performance. Finally, we discuss potential defense methods that can be applied against these attacks.

Securing The Void: Assessing The Dynamic Threat Landscape Of Space, Brianna Bace, Dr. Unal Tatar

Military Cyber Affairs

Outer space is a strategic and multifaceted domain that is a crossroads for political, military, and economic interests. From a defense perspective, the U.S. military and intelligence community rely heavily on satellite networks to meet national security objectives and execute military operations and intelligence gathering. This paper examines the evolving threat landscape of the space sector, encompassing natural and man-made perils, emphasizing the rise of cyber threats and the complexity introduced by dual-use technology and commercialization. It also explores the implications for security and resilience, advocating for collaborative efforts among international organizations, governments, and industry to safeguard the space sector.

Commercial Enablers Of China’S Cyber-Intelligence And Information Operations, Ethan Mansour, Victor Mukora

Military Cyber Affairs

In a globally commercialized information environment, China uses evolving commercial enabler networks to position and project its goals. They do this through cyber, intelligence, and information operations. This paper breaks down the types of commercial enablers and how they are used operationally. It will also address the CCP's strategy to gather and influence foreign and domestic populations throughout cyberspace. Finally, we conclude with recommendations for mitigating the influence of PRC commercial enablers.

Analysis And Evaluation Of Different Transformer Architectures For The Protein Sequence Representation And The Corresponding Hypothetical Applications, Mary Mao

Theses and Dissertations

This project is to investigate and assess several Transformer topologies for the modeling of protein sequences and their corresponding uses. Each major model for the protein sequence representations is inspected with its mathematical theory and analyzed for the different performance of the models with various validation repositories.

Analyzing Information Cascades Through Machine Learning And Data Analytics, Betul Agirman

Honors Scholar Theses

In today's digital age, social media platforms have become pivotal in influencing public opinion and behavior, with information spreading being both beneficial and detrimental. This rapid spread is typically called an information cascade, and they are important in further understanding social influence, managing misinformation, and even predicting potential trends of public responses. With social media, people are connected so easily to one another like a network, wherein it becomes possible for them to influence each other’s behavior and decisions. Utilizing a dataset from Weibo that spans critical periods of the COVID-19 outbreak, this study integrates machine learning and data analytics …

Adaptive Solutions For Improving The Quality Of Mobile User Experiences, Meena V

Theses and Dissertations

Nowadays, mobile applications have drawn a lot of attention as they bring computational and storage resources close to consumers globally through high-speed networks. Applications such as the medical microscope, 2D barcode reader, environmental sensor(s), mobile security and authenticator, vehicle remote controller, and IoT-based synchronizer come pre- applications are resource-intensive, as it performs computation(s) by utilizing diverse services like location, app-tracking, networking, camera, calendar, contacts, Bluetooth, etc. for each user activity. Parallel execution of such intense services might utilize the utmost memory and CPU of the mobile device which in turn degrades the overall Quality of Experience (QoE). This ultimately ends …

Side Channel Detection Of Pc Rootkits Using Nonlinear Phase Space, Rebecca Clark

Poster Presentations

Cyberattacks are increasing in size and scope yearly, and the most effective and common means of attack is through malicious software executed on target devices of interest. Malware threats vary widely in terms of behavior and impact and, thus, effective methods of detection are constantly being sought from the academic research community to offset both volume and complexity. Rootkits are malware that represent a highly feared threat because they can change operating system integrity and alter otherwise normally functioning software. Although normal methods of detection that are based on signatures of known malware code are the standard line of defense, …

High-Altitude Ballooning And Payload Design, Matthew Smith

Honors College Theses

The Icarus project began in 2017 where a team performs high-altitude balloon (HAB) launches to conduct atmospheric and payload design research. The project involves sending a data-collecting payload containing recording equipment to high altitudes (~100,000 feet). This paper outlines the Icarus project as well as the efforts to design a payload that will remain stable throughout flight while also protecting the equipment housed within the payload shell. SOLIDWORKS computer-aided design (CAD) software was used to design a unique payload system. The payload parts were manufactured with PLA filament using a Prusa MK4 3D printer. The secondary goal of this thesis …

Attention Guided Data Augmentation For Improving The Classification Performance Of Vision Transformers., Nada Baili

Electronic Theses and Dissertations

For over a decade, Deep Neural Networks (DNNs) have been rapidly progressing and achieving great success, forming a robust foundation of state of the art machine learning algorithms that impacted various domains. The advances in data acquisition and processing have undeniably played a major role in these breakthroughs. Data is a crucial component in building successful DNNs, as it enables machine learning models to optimize complex architectures, necessary to perform certain difficult tasks. However, acquiring large-scale data sets is not enough to learn robust models with generalizable features. Instead, an ideal training set should be diverse enough and contain enough …