Digital Commons Network^™

Deep Video Understanding With Model Efficiency And Sparse Active Labeling, Aayush Jung Bahadur Rana

Electronic Theses and Dissertations, 2020-

Videos capture the inherently sequential nature of the real world, making automatic video understanding an essential need for automatic understanding of the real world. Due to major advancements in camera, communication, and storage hardware, videos have become a widely used data format for crucial applications such as home automation, security, analysis, robotics, and autonomous driving. Existing methods for video understanding require heavy computation and large training data for good performance, this limits how quick the videos can be processed and how much data can be labeled for training. Real-world video understanding requires analyzing dense scenes and sequential information, which increases …

Studying Memes During Covid Lockdown As A Lens Through Which To Understand Video-Mediated Communication Interactions, Tatyana Claytor

Electronic Theses and Dissertations, 2020-

The purpose of this study is to analyze image macros about video-mediated communication (VMC) created during the time frame of 2020-2021 when people all over the world started using Zoom and VMC for work and school. It is a unique opportunity to study how users' interactions with themselves and with others were affected at a time when a lot of people started using the technology at the same time. Because the focus is on interactions, I narrowed it down to three topics to analyze the memes: presence, self, and space and place to analyze the memes. I chose memes relating …

Annotation Efficient Visual Recognition: From Semi-Supervised To Few-Shot Learning, Mamshad Nayeem Rizve

Electronic Theses and Dissertations, 2020-

In recent years, supervised deep learning has achieved remarkable success in solving a wide range of visual recognition problems. Large-scale labeled datasets have been crucial for this success and the progress has primarily been limited to controlled environments. In this dissertation, we present methods to improve the annotation efficiency of deep visual recognition models and also propose methods to improve the performance of annotation-efficient models in unconstrained open-world settings. To address the annotation bottleneck in supervised learning, we introduce a pseudo-labeling framework for semi-supervised learning. While consistency regularization methods dominate the field, they heavily rely on domain-specific data augmentations, limiting …

Towards Efficient And Effective Representation Learning For Image And Video Understanding, Taojiannan Yang

Electronic Theses and Dissertations, 2020-

Deep learning has achieved tremendous success on various computer vision tasks. However, deep learning methods and models are usually computationally expensive, making it hard to train and deploy, especially on resource-constrained devices. In this dissertation, we explore how to improve the efficiency and effectiveness of deep learning methods from various perspectives. We first propose a new learning method to learn computationally adaptive representations. Traditional neural networks are static. However, our method trains adaptive neural networks that can adjust their computational cost during runtime, avoiding the need to train and deploy multiple networks for dynamic resource budgets. Next, we extend our …

Detecting Team Conflict From Multiparty Dialogue, Ayesha Enayet

Electronic Theses and Dissertations, 2020-

The emergence of online collaboration platforms has dramatically changed the dynamics of human teamwork, creating a veritable army of virtual teams composed of workers in different physical locations. The global world requires a tremendous amount of collaborative problem solving, primarily virtual, making it an excellent domain for computer scientists and team cognition researchers who seek to understand the dynamics involved in collaborative tasks to provide a solution that can support effective collaboration. Mining and analyzing data from collaborative dialogues can yield insights into virtual teams' thought processes and help develop virtual agents to support collaboration. Good communication is indubitably the …

A Study On Robustness And Semantic Understanding Of Visual Models, Madeline Chantry

Electronic Theses and Dissertations, 2020-

Vision models have improved in popularity and performance on many tasks since the emergence of large-scale datasets, improved access to computational resources, and new model architectures like the transformer. However, it is still not well understood if these models can be deployed in the real world. Because these models are "blackbox" architectures, we do not fully understand what these models are truly learning. An understanding of what models learn "underneath the hood" would result in better improvements for real-world scenarios. Motivated by this, we benchmark these impressive visual models using newly proposed datasets and tasks on their robustness and their …

Human Recognition Theory And Facial Recognition Technology: A Topic Modeling Approach To Understanding The Ethical Implication Of A Developing Algorithmic Technologies Landscape On How We View Ourselves And Are Viewed By Others, Hajer Albalawi

Electronic Theses and Dissertations, 2020-

The emergence of algorithmic-driven technology has significantly impacted human life in the current century. Algorithms, as versatile constructs, hold different meanings across various disciplines, including computer science, mathematics, social science, and human-artificial intelligence studies. This study defines algorithms from an ethical perspective as the foundation of an information society and focuses on their implications in the context of human recognition. Facial recognition technology, driven by algorithms, has gained widespread use, raising important ethical questions regarding privacy, bias, and accuracy. This dissertation aims to explore the impact of algorithms on machine perception of human individuals and how humans perceive one another …

Efficient Convolutional Neural Networks For Image Classification And Regression, Muhammad Tayyab

Electronic Theses and Dissertations, 2020-

Neural networks have been a topic of research since 1970s and the Convolutional Neural Networks (CNNs) were first shown to work well for hand written digits recognition in 1998. These early networks were however still shallow and contained only a few layers. Moreover these networks were mostly trained on a small amount of data in contrast to the modern CNNs which contain hundreds of convolution layers and are trained on millions of images. However, this recent shift in machine learning comes at a cost. Modern neural networks have extremely large number of parameters and require huge amount of computations for …

Identification And Modeling Social Media Influence Pathways: A Characterization Of A Disinformation Campaign Using The Flooding-The-Zone Strategy Via Transfer Entropy, Jasser Jasser

Electronic Theses and Dissertations, 2020-

The internet has made it easy for narratives to spread quickly and widely without regard for accuracy or the harm they may cause to society. Unfortunately, this has led to the rise of bad actors who use fake and misleading articles to spread harmful misinformation. These actors flood the information space with low-quality articles in an effort to disrupt opposing narratives, sow confusion, and discourage the pursuit of truth. In societies that prioritize free speech, maintaining control over the information space remains a persistent challenge. Achieving this requires strategic planning to protect the dissemination of information in ways that promote …

Mapping The Focal Points Of Wordpress: A Software And Critical Code Analysis, Bryce Jackson

Electronic Theses and Dissertations, 2020-

Programming languages or code can be examined through numerous analytical lenses. This project is a critical analysis of WordPress, a prevalent web content management system, applying four modes of inquiry. The project draws on theoretical perspectives and areas of study in media, software, platforms, code, language, and power structures. The applied research is based on Critical Code Studies, an interdisciplinary field of study that holds the potential as a theoretical lens and methodological toolkit to understand computational code beyond its function. The project begins with a critical code analysis of WordPress, examining its origins and source code and mapping selected …

Towards Explainable Ai Using Attribution Methods And Image Segmentation, Garrett J. Rocks

Honors Undergraduate Theses

With artificial intelligence (AI) becoming ubiquitous in a broad range of application domains, the opacity of deep learning models remains an obstacle to adaptation within safety-critical systems. Explainable AI (XAI) aims to build trust in AI systems by revealing important inner mechanisms of what has been treated as a black box by human users. This thesis specifically aims to improve the transparency and trustworthiness of deep learning algorithms by combining attribution methods with image segmentation methods. This thesis has the potential to improve the trust and acceptance of AI systems, leading to more responsible and ethical AI applications. An exploratory …

The Effects Of Head-Centric Rest Frames On Egocentric Distance Perception In Virtual Reality, Yahya Hmaiti

Honors Undergraduate Theses

It has been shown through several research investigations that users tend to underestimate distances in virtual reality (VR). Virtual objects that appear close to users wearing a Head-mounted display (HMD) might be located at a farther distance in reality. This discrepancy between the actual distance and the distance observed by users in VR was found to hinder users from benefiting from the full in-VR immersive experience, and several efforts have been directed toward finding the causes and developing tools that mitigate this phenomenon. One hypothesis that stands out in the field of spatial perception is the rest frame hypothesis (RFH), …

Musical Form Reconstruction In Printed And Handwritten Lead Sheets Via Optical Recognition Of Chord Symbols, Nashir A. Janmohamed

Honors Undergraduate Theses

Optical music recognition (OMR) is the field of study which seeks to use computer vision to extract musical information from images. Most OMR work focuses on music symbols (such as notes, time signatures, clefs, etc.); to date, only two prior works pay attention to chord symbols (shorthand notation commonly used in jazz and popular music lead sheets to describe the harmony of the music) in musical documents. Chord symbols lay the foundation for jazz improvisation - a sequence of chord symbols is repeated during the improvisatory section, and the soloist and accompaniment (primarily, though not exclusively) use the chord symbols …

Biomarker Identification For Breast Cancer Types Using Feature Selection And Explainable Ai Methods, David E. La Rosa Giraud

Honors Undergraduate Theses

This paper investigates the impact the LASSO, mRMR, SHAP, and Reinforcement Feature Selection techniques on random forest models for the breast cancer subtypes markers ER, HER2, PR, and TN as well as identifying a small subset of biomarkers that could potentially cause the disease and explain them using explainable AI techniques. This is important because in areas such as healthcare understanding why the model makes a specific decision is important it is a diagnostic of an individual which requires reliable AI. Another contribution is using feature selection methods to identify a small subset of biomarkers capable of predicting if a …

Understanding, Modeling, And Simulating The Discrepancy Between Intended And Perceived Image Appearance On Optical See-Through Augmented Reality Displays, Austin Erickson

Electronic Theses and Dissertations, 2020-

Augmented reality (AR) displays are transitioning from being primarily used in research and development settings, to being used by the general public. With this transition, these displays will be used by more people, in many different environments, and in many different contexts. Like other displays, the user's perception of virtual imagery is influenced by the characteristics of the user's environment, creating a discrepancy between the intended appearance and the perceived appearance of virtual imagery shown on the display. However, this problem is much more apparent for optical see-through AR displays, such as the HoloLens. For these displays, imagery is superimposed …

From Human Behavior To Machine Behavior, Zerong Xi

Electronic Theses and Dissertations, 2020-

A core pursuit of artificial intelligence is the comprehension of human behavior. Imbuing intelligent agents with a good human behavior model can help them understand how to behave intelligently and interactively in complex situations. Due to the increase in data availability and computational resources, the development of machine learning algorithms for duplicating human cognitive abilities has made rapid progress. To solve difficult scenarios, learning-based methods must search for solutions in a predefined but large space. Along with implementing a smart exploration strategy, the right representation for a task can help narrow the search process during learning. This dissertation tackles three …

Improving Deep Neural Network Training With Knowledge Distillation, Dongdong Wang

Electronic Theses and Dissertations, 2020-

Knowledge distillation, as a popular compression technique, has been widely used to reduce deep neural network (DNN) size for a variety of applications. However, in recent years, some research had found its potential for improving deep neural network performance. This dissertation focuses on further exploring its power to facilitate accurate and reliable DNN training. First, I explored data-efficient method for blackbox knowledge distillation where the specifics of the DNN for distillation is inaccessible. I integrated active learning and mixup to obtain significant distillation performance gain with limited data. This work reveals the competence of knowledge distillation to facilitate large foundation …

Towards Optimization And Robustification Of Data-Driven Models, Ehsan Kazemi Foroushani

Electronic Theses and Dissertations, 2020-

In the past two decades, data-driven models have experienced a renaissance, with notable success achieved through the use of models such as deep neural networks (DNNs) in various applications. However, complete reliance on intelligent machine learning systems is still a distant dream. Nevertheless, the initial success of data-driven approaches presents a promising path for building trustworthy data-oriented models. This thesis aims to take a few steps toward improving the performance of existing data-driven frameworks in both the training and testing phases. Specifically, we focus on several key questions: 1) How to efficiently design optimization methods for learning algorithms that can …

Deep Learning Approaches For Automatic Colorization, Super-Resolution, And Representation Of Volumetric Data, Sudarshan Devkota

Graduate Thesis and Dissertation 2023-2024

This dissertation includes a collection of studies that aim to improve the way we represent and visualize volume data. The advancement of medical imaging has revolutionized healthcare, providing crucial anatomical insights for accurate diagnosis and treatment planning. Our first study introduces an innovative technique to enhance the utility of medical images, transitioning from monochromatic scans to vivid 3D representations. It presents a framework for reference-based automatic color transfer, establishing deep semantic correspondences between a colored reference image and grayscale medical scans. This methodology extends to volumetric rendering, eliminating the need for manual intervention in parameter tuning. Next, it delves into …

A Systematic Review Of Cryptocurrencies Use In Cybercrimes, Kieran B D Human

Graduate Thesis and Dissertation 2023-2024

Cryptocurrencies are one of the most prominent applications of blockchain systems. While cryptocurrencies promise many features and advantages, such as decentralization, anonymity, and ease of access, those very features can be abused. For instance, as documented in various recent works, cryptocurrencies have been frequently abused in many different forms of cybercrime. Despite the plethora of works on measuring and understanding the abuse of cryptocurrencies in the digital space, there has been no work on systemizing this knowledge by comprehensively understanding those contributions, contrasting them based on their merit, and understanding the gap in this research space.

This thesis initiates the …

Exploring The Feasibility Of Machine Learning Techniques In Recognizing Complex Human Activities, Shengnan Hu

Graduate Thesis and Dissertation 2023-2024

This dissertation introduces several technical innovations that improve the ability of machine learning models to recognize a wide range of complex human activities. As human sensor data becomes more abundant, the need to develop algorithms for understanding and interpreting complex human actions has become increasingly important. Our research focuses on three key areas: multi-agent activity recognition, multi-person pose estimation, and multimodal fusion.

To tackle the problem of monitoring coordinated team activities from spatio-temporal traces, we introduce a new framework that incorporates field of view data to predict team performance. Our framework uses Spatial Temporal Graph Convolutional Networks (ST-GCN) and recurrent …

Material Appearance Modeling For Physically Based Rendering, Alexis Benamira

Graduate Thesis and Dissertation 2023-2024

Photorealistic rendering focuses on creating images with a computer that imitates pictures of reallife scenes as faithfully as possible. To achieve this, rendering algorithms require incorporating accurate modeling of how light interacts with various types of matter. For most objects, this model needs to account for the scattering of the light rays. However, this model falls short when rendering objects of sizes smaller or comparable to the wavelength of the incident light. In this case, new phenomena such as diffraction or interference are observed and have been characterized in optics. Digital rendering of those phenomena involve different light representations than …

Optimizing Deep Neural Networks Performance: Efficient Techniques For Training And Inference, Ankit Sharma

Graduate Thesis and Dissertation 2023-2024

Recent advances in computer vision tasks are mainly due to the success of large deep neural networks. The current state-of-the-art models have high computational costs during inference and suffer from a high memory footprint. Therefore, deploying these large networks on edge devices remains a serious concern. Furthermore, training these over-parameterized networks is computationally expensive and requires a longer training time. Thus, there is a demand to develop techniques that can efficiently reduce training costs and also be able to deploy neural networks on mobile and embedded devices. This dissertation presents practices like designing a lightweight network architecture and increasing network …

Reconstructing 3d Humans From Visual Data, Ce Zheng

Graduate Thesis and Dissertation 2023-2024

Understanding humans in visual content is fundamental for numerous computer vision applications. Extensive research has been conducted in the field of human pose estimation (HPE) to accurately locate joints and construct body representations from images and videos. Expanding on HPE, human mesh recovery (HMR) addresses the more complex task of estimating the 3D pose and shape of the entire human body. HPE and HMR have gained significant attention due to their applications in areas such as digital human avatar modeling, AI coaching, and virtual reality [135]. However, HPE and HMR come with notable challenges, including intricate body articulation, occlusion, depth …

Towards A Robust And Efficient Deep Neural Network For The Lidar Point Cloud Perception, Zixiang Zhou

Graduate Thesis and Dissertation 2023-2024

In recent years, LiDAR has emerged as a crucial perception tool for robotics and autonomous vehicles. However, most LiDAR perception methods are adapted from 2D image-based deep learning methods, which are not well-suited to the unique geometric structure of LiDAR point cloud data. This domain gap poses challenges for the fast-growing LiDAR perception tasks. This dissertation aims to investigate suitable deep network structures tailored for LiDAR point cloud data, and therefore design a more efficient and robust LiDAR perception framework. Our approach to address this challenge is twofold. First, we recognize that LiDAR point cloud data is characterized by an …

Towards A Holistic And Comparative Analysis Of The Free Content Web: Security, Privacy, And Performance, Abdulrahman Alabduljabbar

Electronic Theses and Dissertations, 2020-

Free content websites that provide free books, music, games, movies, etc., have existed on the Internet for many years. While it is a common belief that such websites might be different from premium websites providing the same content types in terms of their security, a rigorous analysis that supports this belief is lacking from the literature. In particular, it is unclear if those websites are as safe as their premium counterparts. In this dissertation, we set out to investigate the similarities and differences between free content and premium websites, including their risk profiles. Moreover, we analyze and quantify through measurements …

Machine Learning Algorithms For Molecular Signature Identification With High-Throughput Genome Sequencing Data, Jiao Sun

Electronic Theses and Dissertations, 2020-

Powered by the high-throughput genomic technologies, the RNA sequencing (RNA-Seq) method is capable of measuring transcriptome-wide mRNA expressions and molecular activities in cells. Elucidation of gene expressions at the isoform resolution enables the detection of better molecular signatures for phenotype prediction, and the identified biomarkers may provide insights into the functional consequences of disease. This dissertation research focuses on developing advanced machine learning algorithms for mining large-scale RNA-Seq data in cancer transcriptome analysis. A platform-integrated model for transcript quantification (IntMTQ) is developed to improve the performance of RNA-Seq on isoform expression estimation. IntMTQ provides more precise RNA-Seq-based isoform quantification, and …

Methodologies For Evaluating Interaction Cues For Virtual Reality, Xinyu Hu

Electronic Theses and Dissertations, 2020-

Virtual reality (VR) games and educational systems commonly employ interaction cues to provide information on how to take appropriate actions at particular moments. Interaction cues can be employed for different purposes, such as informing the user to look, go, pick, and operate. Additionally, different types of interaction cues can directly affect usability and user experiences. In our early research, we conducted two ecologically valid empirical studies with a preexisting VR training application and evaluated the effects of delayed interaction cues, in addition to comparing the purposes of interaction cues for learning and retention. Our results indicated that immediate interaction cues …

Low-Resource Machine Learning Techniques For The Analysis Of Online Social Media Textual Data, Toktam Amanzadeh Oghaz

Electronic Theses and Dissertations, 2020-

Low-resource and label-efficient machine learning methods can be described as the family of statistical and machine learning techniques that can achieve high performance without needing a substantial amount of labeled data. These methods include both unsupervised learning techniques, such as LDA, and supervised methods, such as active learning, each providing different benefits. Thus, this dissertation is devoted to the design and analysis of unsupervised and supervised techniques to provide solutions for the following problems: Unsupervised narrative summary extraction for social media content, Social media text classification with Active Learning (AL), Investigating restrictions and benefits of using Curriculum Learning (CL) for …

Towards More Efficient Collaborative Distributed Data Analysis And Learning, Zixia Liu

Electronic Theses and Dissertations, 2020-

Modern information era gives rise to the persistent generation of large amounts of data with rapid speed and broad geographical distribution. Obtaining knowledge and understanding via analysis and learning from such data have invaluable worth. Features of such data analytical tasks commonly include: data can be large scale and geographically distributed; computing capability demand can be enormous; tasks can be time-critical; some data can be private; participants can have heterogeneous capabilities and non-IID data; and multiple simultaneously submitted data analytical tasks can be possible. These bring challenges to contemporary computing infrastructure and learning models. In view of this, we develop …