Computer Sciences Commons^™

Increasing The Robustness Of Machine Learning By Adversarial Attacks, Gourab Mukhopadhyay

Theses and Dissertations

By perturbation or physical attacks any machine can be fooled into predicting something else other than the intended output. There are training data based on which the model is trained to predict unknown things. The objective was to create noises and shades of different levels on the images and do experiments for measuring accuracy and making the model classify the traffic signs. When it comes to adding shades to the pictures, pixels were modified for three different layers of the pictures. The experiment also shows that with the shadows getting deeper, the accuracies drop significantly. Here, some changes in pixels …

Personalized Driving Using Inverse Reinforcement Learning, Rodrigo J. Gonzalez Salinas

Theses and Dissertations

This thesis introduces an autonomous driving controller designed to replicate individual driving behaviors based on a provided demonstration. The controller employs Inverse Reinforcement Learning (IRL) to formulate the reward function associated with the provided demonstration. IRL is implemented through a dual-feedback loop system. The inner loop utilizes Q-learning, a model-free reinforcement learning technique, to optimize the Hamilton-Jacobi-Bellman (HJB) equation and derive an appropriate control solution. The outer loop leverages this derived control solution to generate parameters for the reward function, which are subsequently integrated into the HJB equation. The ultimate control policy is deduced from the final reward function obtained …

Enhancing Tumor Classification Through Machine Learning Algorithms For Breast Cancer Diagnosis, Lawrence Agbota, Edmund Agyemang, Priscilla Kissi-Appiah, Lateef Moshood, Akua Osei- Nkwantabisa, Vincent Agbenyeavu, Abraham Nsiah, Augustina Adjei

School of Mathematical and Statistical Sciences Faculty Publications and Presentations

In cancer diagnosis, machine learning helps improve cancer detection by providing doctors with a second perspective and allowing for faster and more accurate determination and decisions. Numerous studies have used both classic machine learning approaches and deep learning to address cancer classification. In this study, we examine the efficacy of five commonly used machine learning algorithms; both traditional and deep learning models namely, Logistic Regression, Support Vector Machines (SVM), Random Forest (RF), Decision Tree and Deep Neural Networks (DNN). We analyze their ability to properly classify tumors as Benign or Malignant using the Wisconsin breast cancer dataset (WBCD). Random Forest …

Methionine Sulfoxide Speciation In Mouse Hippocampus Revealed By Global Proteomics Exhibits Age- And Alzheimer’S Disease-Dependent Changes Targeted To Mitochondrial And Glycolytic Pathways, Filipa Blasco Tavares Pereira Lopes, Daniela Schlatzer, Mengzhen Li, Serhan Yılmaz, Rihua Wang, Xin Qi, Marzieh Ayati, Mehmet Koyutürk, Mark R. Chance

Computer Science Faculty Publications and Presentations

Methionine oxidation to the sulfoxide form (MSox) is a poorly understood post-translational modification of proteins associated with non-specific chemical oxidation from reactive oxygen species (ROS), whose chemistries are linked to various disease pathologies, including neurodegeneration. Emerging evidence shows MSox site occupancy is, in some cases, under enzymatic regulatory control, mediating cellular signaling, including phosphorylation and/or calcium signaling, and raising questions as to the speciation and functional nature of MSox across the proteome. The 5XFAD lineage of the C57BL/6 mouse has well-defined Alzheimer’s and aging states. Using this model, we analyzed age-, sex-, and disease-dependent MSox speciation in the mouse hippocampus. …

Methionine Sulfoxide Speciation In Mouse Hippocampus Revealed By Global Proteomics Exhibits Age- And Alzheimer’S Disease-Dependent Changes Targeted To Mitochondrial And Glycolytic Pathways, Filipa Blasco Tavares Pereira Lopes, Daniela Schlatzer, Mengzhen Li, Serhan Yılmaz, Rihua Wang, Xin Qi, Marzieh Ayati, Mehmet Koyutürk, Mark R. Chance

Computer Science Faculty Publications and Presentations

Methionine oxidation to the sulfoxide form (MSox) is a poorly understood post-translational modification of proteins associated with non-specific chemical oxidation from reactive oxygen species (ROS), whose chemistries are linked to various disease pathologies, including neurodegeneration. Emerging evidence shows MSox site occupancy is, in some cases, under enzymatic regulatory control, mediating cellular signaling, including phosphorylation and/or calcium signaling, and raising questions as to the speciation and functional nature of MSox across the proteome. The 5XFAD lineage of the C57BL/6 mouse has well-defined Alzheimer’s and aging states. Using this model, we analyzed age-, sex-, and disease-dependent MSox speciation in the mouse hippocampus. …

Exploring Graph Neural Networks In Reinforcement Learning: A Comparative Study On Architectures For Locomotion Tasks, Gaukhar Nurbek

Theses and Dissertations

Deep Reinforcement learning (DRL) has gained importance in optimizing control policies, while Graph Neural Networks (GNNs) offer a robust approach for modeling complex relationships within systems represented as graphs. This thesis investigates the integration of DRL and GNNs to optimize control policies for robotic tasks, with a focus on locomotion. It compares static and dynamic GNN architectures for control policy predictions, revealing their strengths and limitations in adapting to locomotion predictions. The study assesses the impact of model structure complexity on GNNs' predictive capabilities, showcasing how intricate model structure can maximize GNNs' potential in capturing spatial and relational dependencies when …

Interpretable Spatio-Temporal Embedding For Brain Structural-Effective Network With Ordinary Differential Equation, Haoteng Tang, Guodong Liu, Siyuan Dai, Kai Ye, Kun Zhao, Wenlu Wang, Carl Yang, Lifang He, Alex D. Leow, Paul Thompson

Computer Science Faculty Publications and Presentations

The MRI-derived brain network serves as a pivotal instrument in elucidating both the structural and functional aspects of the brain, encompassing the ramifications of diseases and developmental processes. However, prevailing methodologies, often focusing on synchronous BOLD signals from functional MRI (fMRI), may not capture directional influences among brain regions and rarely tackle temporal functional dynamics. In this study, we first construct the brain-effective network via the dynamic causal model. Subsequently, we introduce an interpretable graph learning framework termed Spatio-Temporal Embedding ODE (STE-ODE). This framework incorporates specifically designed directed node embedding layers, aiming at capturing the dynamic interplay between structural and …

Care-Teach: Proposing An Open-Source Approach To Personalized Learning, Jaime Augusto Alvarez Perez

Theses and Dissertations

Care-Teach is an algorithmic educational model that provides teachers and educators with the tools to create interactive, text-based lessons that address a student’s need for continuity and reinforcement. Care-Teach is built around two core components: a Student Behavior Profile and the Skill Tree. These models work together to give each student a personalized learning experience that reinforces their pre-existing strengths and inclinations. The Student Behavior Profile model keeps track of a learner’s inclinations and mood, utilizing metrics such as average response time and accuracy to categorize opportunities for educator involvement. The Skill Tree is an organizational …

Mechanistic Investigation Of C—C Bond Activation Of Phosphaalkynes With Pt(0) Complexes, Roberto M. Escobar, Abdurrahman C. Ateşin, Christian Müller, William D. Jones, Tülay Ateşin

Research Symposium

Carbon–carbon (C–C) bond activation has gained increased attention as a direct method for the synthesis of pharmaceuticals. Due to the thermodynamic stability and kinetic inaccessibility of the C–C bonds, however, activation of C–C bonds by homogeneous transition-metal catalysts under mild homogeneous conditions is still a challenge. Most of the systems in which the activation occurs either have aromatization or relief of ring strain as the primary driving force. The activation of unstrained C–C bonds of phosphaalkynes does not have this advantage. This study employs Density Functional Theory (DFT) calculations to elucidate Pt(0)-mediated C–CP bond activation mechanisms in phosphaalkynes. Investigating the …

Ex-Vivo Hippocampus Segmentation Using Diffusion-Weighted Mri, Haoteng Tang, Siyuan Dai, Eric M. Zou, Guodong Liu, Ryan Ahearn, Ryan Krafty, Michel Modo, Liang Zhan

Computer Science Faculty Publications and Presentations

The hippocampus is a crucial brain structure involved in memory formation, spatial navigation, emotional regulation, and learning. An accurate MRI image segmentation of the human hippocampus plays an important role in multiple neuro-imaging research and clinical practice, such as diagnosing neurological diseases and guiding surgical interventions. While most hippocampus segmentation studies focus on using T1-weighted or T2-weighted MRI scans, we explore the use of diffusion-weighted MRI (dMRI), which offers unique insights into the microstructural properties of the hippocampus. Particularly, we utilize various anisotropy measures derived from diffusion MRI (dMRI), including fractional anisotropy, mean diffusivity, axial diffusivity, and radial diffusivity, for …

Possible Role Of Correlation Coefficients And Network Analysis Of Multiple Intracellular Proteins In Blood Cells Of Patients With Bipolar Disorder In Studying The Mechanism Of Lithium Responsiveness: A Proof-Concept Study, Keming Gao, Marzieh Ayati, Nicholas M. Kaye, Mehmet Koyutürk, Joseph R. Calabrese, Eric Christian, Hillard M. Lazarus, David Kaplan

Computer Science Faculty Publications and Presentations

Background: The mechanism of lithium treatment responsiveness in bipolar disorder (BD) remains unclear. The aim of this study was to explore the utility of correlation coefficients and protein-to-protein interaction (PPI) network analyses of intracellular proteins in monocytes and CD4+ lymphocytes of patients with BD in studying the potential mechanism of lithium treatment responsiveness. Methods: Patients with bipolar I or II disorder who were diagnosed with the MINI for DSM-5 and at any phase of the illness with at least mild symptom severity and received lithium (serum level ≥ 0.6 mEq/L) for 16 weeks were divided into two groups, responders (≥50% …

Machine Learning For Wireless Network Throughput Prediction, Gustavo A. Fernandez

School of Mathematical and Statistical Sciences Faculty Publications and Presentations

This paper analyzes a dataset containing radio frequency (RF) measurements and Key Performance Indicators (KPIs) captured at 1876.6MHz with a bandwidth of 10MHz from an operational 4G LTE network in Nigeria. The dataset includes metrics such as RSRP (Reference Signal Received Power), which measures the power level of reference signals; RSRQ (Reference Signal Received Quality), an indicator of signal quality that provides insight into the number of users sharing the same resources; RSSI (Received Signal Strength Indicator), which gauges the total received power in a bandwidth; SINR (Signal to Interference plus Noise Ratio), a measure of signal quality considering both …

Efficient High-Resolution Time Series Classification Via Attention Kronecker Decomposition, Aosong Feng, Jialin Chen, Juan Garza, Brooklyn Berry, Francisco Salazar, Yifeng Gao, Rex Ying, Leandros Tassiulas

Computer Science Faculty Publications and Presentations

The high-resolution time series classification problem is essential due to the increasing availability of detailed temporal data in various domains. To tackle this challenge effectively, it is imperative that the state-of-theart attention model is scalable to accommodate the growing sequence lengths typically encountered in highresolution time series data, while also demonstrating robustness in handling the inherent noise prevalent in such datasets. To address this, we propose to hierarchically encode the long time series into multiple levels based on the interaction ranges. By capturing relationships at different levels, we can build more robust, expressive, and efficient models that are capable of …

Constrained Multiview Representation For Self-Supervised Contrastive Learning, Siyuan Dai, Kai Ye, Kun Zhao, Ge Cui, Haoteng Tang, Liang Zhan

Computer Science Faculty Publications and Presentations

Representation learning constitutes a pivotal cornerstone in contemporary deep learning paradigms, offering a conduit to elucidate distinctive features within the latent space and interpret the deep models. Nevertheless, the inherent complexity of anatomical patterns and the random nature of lesion distribution in medical image segmentation pose significant challenges to the disentanglement of representations and the understanding of salient features. Methods guided by the maximization of mutual information, particularly within the framework of contrastive learning, have demonstrated remarkable success and superiority in decoupling densely intertwined representations. However, the effectiveness of contrastive learning highly depends on the quality of the positive and …

A Simple Proof That Ricochet Robots Is Pspace-Complete, Jose Balanza-Martinez, Angel A. Cantu, Robert Schweller, Tim Wylie

Computer Science Faculty Publications and Presentations

In this paper, we seek to provide a simpler proof that the relocation problem in Ricochet Robots (Lunar Lockout with fixed geometry) is PSPACE-complete via a reduction from Finite Function Generation (FFG). Although this result was originally proven in 2003, we give a simpler reduction by utilizing the FFG problem, and put the result in context with recent publications showing that relocation is also PSPACE-complete in related models.

Self Pre-Training With Topology- And Spatiality-Aware Masked Autoencoders For 3d Medical Image Segmentation, Pengfei Gu, Yejia Zhang, Huimin Li, Chaoli Wang, Danny Z. Chen

Computer Science Faculty Publications and Presentations

Masked Autoencoders (MAEs) have been shown to be effective in pre-training Vision Transformers (ViTs) for natural and medical image analysis problems. By reconstructing missing pixel/voxel information in visible patches, a ViT encoder can aggregate contextual information for downstream tasks. But, existing MAE pre-training methods, which were specifically developed with the ViT architecture, lack the ability to capture geometric shape and spatial information, which is critical for medical image segmentation tasks. In this paper, we propose a novel extension of known MAEs for self pre-training (i.e., models pre-trained on the same target dataset) for 3D medical image segmentation. (1) We propose …

Fledge: Ledger-Based Federated Learning Resilient To Inference And Backdoor Attacks, Jorge Castillo, Phillip Rieger, Hossein Fereidooni, Qian Chen, Ahmad Sadeghi

Informatics and Engineering Systems Faculty Publications and Presentations

Federated learning (FL) is a distributed learning process that uses a trusted aggregation server to allow multiple parties (or clients) to collaboratively train a machine learning model without having them share their private data. Recent research, however, has demonstrated the effectiveness of inference and poisoning attacks on FL. Mitigating both attacks simultaneously is very challenging. State-of-the-art solutions have proposed the use of poisoning defenses with Secure Multi-Party Computation (SMPC) and/or Differential Privacy (DP). However, these techniques are not efficient and fail to address the malicious intent behind the attacks, i.e., adversaries (curious servers and/or compromised clients) seek to exploit a …

Multitasking Scheduling With Shared Processing, Bin Fu, Yumei Huo, Hairong Zhao

Computer Science Faculty Publications and Presentations

Recently, the problem of multitasking scheduling has raised a lot of interest in the service industries. Hall et al. (Discrete Applied Mathematics, 2016) proposed a shared processing multitasking scheduling model which allows a team to continue to work on the primary tasks while processing the routinely scheduled activities as they occur. With a team being modeled as a single machine, the processing sharing of the machine is achieved by allocating a fraction of the processing capacity to routine jobs and the remaining fraction, which we denote as sharing ratio, to the primary jobs. In this paper, we generalize this model …

Private Ethereum Blockchain Implementation And Its Security Features For Smart Home Iot, Hasibul Grande Alam

Theses and Dissertations

The security and privacy of IoT devices have become primary concerns as smart home networks are connected to the internet. Ethereum blockchain can be a solution to mitigate or prevent attacks – sniffing attacks, malware attacks, Eavesdropping, and Distributed Denial of Services (DDoS) attacks. Deploying Ethereum in resource constraint IoT devices is challenging due to resultant energy consumption, computational overhead, and delay. We adopted smart home as a case study to examine our methodology as a model for general IoT applications. This thesis work presents the implementation of private Ethereum blockchain that is optimized and installable on smart home IoT. …

Intellibeehive, Christian Ivan Narcia-Macias

Theses and Dissertations

Utilizing computer vision and the latest technological advancements, in this study, we developed a honey bee monitoring system that aims to enhance our understanding of Colony Collapse Disorder, honey bee behavior, population decline, and overall hive health. The system is positioned at the hive entrance providing real-time data, enabling beekeepers to closely monitor the hive's activity and health through an account-based website. Using machine learning, our monitoring system can accurately track honey bees, monitor pollen-gathering activity, and detect Varroa mites, all without causing any disruption to the honey bees. Moreover, we have ensured that the development of this monitoring system …

Enhancing Time Series Hashing Performance Via Deep Orthogonal Hashing, Mahmudul Hasan Robin

Theses and Dissertations

Deep hashing has been widely used for efficient retrieval and classification of high-dimensional data like images and text. However, its application to time series data is still challenging due to the data’s temporal nature. To tackle this issue, a new deep hashing method has been proposed that generates efficient hash codes and enhances the time series hashing performance using a ResNet model with Orthohash (Cosine Similarity Loss). The proposed method uses one loss architecture while using ResNet model for efficient hashing. It uses the Character Trajectories dataset to extract discriminative features from the time series data. These features are then …

Robust And Uncertainty-Aware Image Classification Using Bayesian Vision Transformer Model, Fazlur Rahman Bin Karim

Theses and Dissertations

Transformer Neural Networks have emerged as the predominant architecture for addressing a wide range of Natural Language Processing (NLP) applications such as machine translation, speech recognition, sentiment analysis, text anomaly detection, etc. This noteworthy achievement of Transformer Neural Networks in the NLP field has sparked a growing interest in integrating and utilizing Transformer models in computer vision tasks. The Vision Transformer (ViT) model efficiently captures long-range dependencies by employing a self-attention mechanism to transform different image data into meaningful, significant representations. Recently, the Vision Transformer (ViT) has exhibited incredible performance in solving image classification problems by utilizing ViT models, thereby …

A Fast And Responsive Web-Based Framework For Visualizing Hpc Application Usage, Ved Arora, Nayeli Gurrola, Amiya K. Maji, Guangzhen Jin

Computer Science Faculty Publications and Presentations

Insights about applications and user environments can help HPC center staff make data-driven decisions about cluster operations. In this paper, we present a fast and responsive web-based visualization framework for analyzing HPC application usage. By leveraging XALT, a powerful tool for tracking application and library usage, we collected tens of millions of data points on a national supercomputer. The portable visualization framework created with Plotly Dash can be easily launched as a container and accessed from a web browser. The presented visualizations take a deep dive into the XALT data, analyzing application use, compiler usage, library usage, and even user-specific …

A Comprehensive Survey Of Complex Brain Network Representation, Haoteng Tang, Guixiang Ma, Yanfu Zhang, Kai Ye, Lei Guo, Guodong Liu, Qi Huang, Yalin Wang, Olusola Ajilore, Alex D. Leow

Computer Science Faculty Publications and Presentations

Highlights

• Major traditional and deep learning methods on brain network representation are overviewed.

• Brain network datasets and algorithm implementation tools are summarized.

• Promising research directions in brain network analysis are discussed.

Abstract

Recent years have shown great merits in utilizing neuroimaging data to understand brain structural and functional changes, as well as its relationship to different neurodegenerative diseases and other clinical phenotypes. Brain networks, derived from different neuroimaging modalities, have attracted increasing attention due to their potential to gain system-level insights to characterize brain dynamics and abnormalities in neurological conditions. Traditional methods aim to pre-define multiple topological features of brain …

A Review Of Cyber Attacks On Sensors And Perception Systems In Autonomous Vehicle, Taminul Islam, Md. Alif Sheakh, Anjuman Naher Jui, Omar Sharif, Md Zobaer Hasan

School of Mathematical and Statistical Sciences Faculty Publications and Presentations

Vehicle automation has been in the works for a long time now. Automatic brakes, cruise control, GPS satellite navigation, etc. are all common features seen in today's automobiles. Automation and artificial intelligence breakthroughs are likely to lead to an increase in the usage of automation technologies in cars. Because of this, mankind will be more reliant on computer-controlled equipment and car systems in our daily lives. All major corporations have begun investing in the development of self-driving cars because of the rapid advancement of advanced driver support technologies. However, the level of safety and trustworthiness is still questionable. Imagine what …

Sublinear Time Motif Discovery From Multiple Sequences, Bin Fu, Yunhui Fu, Yuan Xue

Computer Science Faculty Publications and Presentations

In this paper, a natural probabilistic model for motif discovery has been used to experimentally test the quality of motif discovery programs. In this model, there are k background sequences, and each character in a background sequence is a random character from an alphabet, Σ. A motif G = g1g2 . . . gm is a string of m characters. In each background sequence is implanted a probabilistically-generated approximate copy of G. For a probabilistically-generated approximate copy b1b2 . . . bm of G, every character, bi , is probabilistically generated, such that the probability for bi 6= gi is …

Complexity Of Reconfiguration In Surface Chemical Reaction Networks, Robert M. Alaniz, Josh Brunner, Michael Coulombe, Erik D. Demaine, Yevhenii Diomidov, Ryan Knobel, Timothy Gomez, Elise Grizzell, Jayson Lynch, Andrew Rodriguez, Robert Schweller, Tim Wylie

Computer Science Faculty Publications and Presentations

We analyze the computational complexity of basic reconfiguration problems for the recently introduced surface Chemical Reaction Networks (sCRNs), where ordered pairs of adjacent species nondeterministically transform into a different ordered pair of species according to a predefined set of allowed transition rules (chemical reactions). In particular, two questions that are fundamental to the simulation of sCRNs are whether a given configuration of molecules can ever transform into another given configuration, and whether a given cell can ever contain a given species, given a set of transition rules. We show that these problems can be solved in polynomial time, are NP-complete, …

Invading The Integrity Of Deep Learning (Dl) Models Using Lsb Perturbation & Pixel Manipulation, Ashraful Tauhid

Theses and Dissertations

The use of deep learning (DL) models for solving classification and recognition-related problems are expanding at an exponential rate. However, these models are computationally expensive both in terms of time and resources. This imposes an entry barrier for low-profile businesses and scientific research projects with limited resources. Therefore, many organizations prefer to use fully outsourced trained models, cloud computing services, pre-trained models are available for download and transfer learning. This ubiquitous adoption of DL has unlocked numerous opportunities but has also brought forth potential threats to its prospects. Among the security threats, backdoor attacks and adversarial attacks have emerged as …

Fedbiometric: Image Features Based Biometric Presentation Attack Detection Using Hybrid Cnns-Svm In Federated Learning, S M Sarwar

Theses and Dissertations

In the past few years, biometric identification systems have become popular for personal, national, and global security. In addition to other biometric modalities, facial and fingerprint recognition have gained popularity due to their uniqueness, stability, convenience, and cost-effectiveness compared to other biometric modalities. However, the evolution of fake biometrics, such as printed materials, 2D or 3D faces, makeup, and cosmetics, has brought new challenges. As a result of these modifications, several facial and fingerprint Presentation Attack Detection methods have been proposed to distinguish between live and spoof faces or fingerprints. Federated learning can play a significant role in this problem …

Simulating Motion Success With Muscle Deficiency In A Musculoskeletal Model Using Reinforcement Learning, Daniel Castillo

Theses and Dissertations

Humans possess an extraordinary ability to execute complex movements, captivating the attention of researchers who strive to develop methods for simulating these actions within a physics-based environment. Motion Capture data stands out as a crucial tool among the proven approaches to tackle this challenge. In this research, we explore the effects of decreased muscle force on the body's capacity to perform various tasks, ranging from simple walking to executing complex jumping jacks. Through a systematic reduction of the allowed force applied to individual muscles or muscle groups, we aim to identify the threshold at which the body's muscles tolerate the …