Digital Commons Network^™

Applications Of Machine Learning To Threat Intelligence, Intrusion Detection And Malware, Charity Barker

Senior Honors Theses

Artificial Intelligence (AI) and Machine Learning (ML) are emerging technologies with applications to many fields. This paper is a survey of use cases of ML for threat intelligence, intrusion detection, and malware analysis and detection. Threat intelligence, especially attack attribution, can benefit from the use of ML classification. False positives from rule-based intrusion detection systems can be reduced with the use of ML models. Malware analysis and classification can be made easier by developing ML frameworks to distill similarities between the malicious programs. Adversarial machine learning will also be discussed, because while ML can be used to solve problems or …

Text Analytics, Nlp, And Accounting Research, Richard M. Crowley

Research Collection School Of Accountancy

The presentation covered: What is text analytics and NLP?; How text analytics has evolved in the accounting literature since the 1980s; What current (as of 2020) methods are used in the literature; What methods are on the horizon.

Does Applying Deep Learning In Financial Sentiment Analysis Lead To Better Classification Performance?, Tao Wang, Changhe Yuan, Cuiyuan Wang

Publications and Research

Using a unique data set from Seeking Alpha, we compare the deep learning approach with traditional machine learning approaches in classifying financial text. We apply the long short-term memory (LSTM) as the deep learning method and Naive Bayes, SVM, Logistic Regression, XGBoost as the traditional machine learning approaches. The results suggest that the LSTM model outperforms the conventional machine learning methods on all metrics. Based on the tSNE graph, the success of the LSTM model is partially explained as the high-accuracy LSTM model distinguishes between positive and negative important sentiment words while those words are chosen based on SHAP values …

High Performance And Machine Learning Algorithms For Brain Fmri Data, Taban Eslami

Dissertations

Brain disorders are very difficult to diagnose for reasons such as overlapping nature of symptoms, individual differences in brain structure, lack of medical tests and unknown causes of some disorders. The current psychiatric diagnostic process is based on behavioral observation and may be prone to misdiagnosis.

Noninvasive brain imaging technologies such as Magnetic Resonance Imaging (MRI) and functional Magnetic Resonance Imaging (fMRI) make the process of understanding the structure and function of the brain easier. Quantitative analysis of brain imaging data using machine learning and data mining techniques can be advantageous not only to increase the accuracy of brain disorder …

Integration Of Forecasting, Scheduling, Machine Learning, And Efficiency Improvement Methods Into The Sport Management Industry, Caleb Mcgrady

Senior Honors Theses

Sport management is a complicated and economically impactful industry and involves many crucial decisions: such as which players to retain or release, how many concession vendors to add, how many fans to expect, what teams to schedule, and many others are made each offseason and changed frequently. The task of making such decisions effectively is difficult, but the process can be made easier using methods of industrial and systems engineering (ISE). Integrating methods such as forecasting, scheduling, machine learning, and efficiency improvement from ISE can be revolutionary in helping sports organizations and franchises be consistently successful. Research shows areas including …

Data Representation For Motor Imagery Classification, Andrew Festa

Theses

While much progress has been made towards the advancement of brain-controlled interfaces (BCI), there remains an information gap between the various domains involved in progressing this area of research. Thus, this research seeks to address this gap through creation of a method of representing brainwave signals in a manner that is intuitive and easy to interpret for both neuroscientists and computer scientists. This method of data representation was evaluated on the ability of the model to accurately classify motor imagery events in a timely manner.

The proposed data representation of electroencephalographic signals in the form of signal images was found …

Truck Trailer Classification Using Side-Fire Light Detection And Ranging (Lidar) Data, Olcay Sahin

Civil & Environmental Engineering Theses & Dissertations

Classification of vehicles into distinct groups is critical for many applications, including freight and commodity flow modeling, pavement management and design, tolling, air quality monitoring, and intelligent transportation systems. The Federal Highway Administration (FHWA) developed a standardized 13-category vehicle classification ruleset, which meets the needs of many traffic data user applications. However, some applications need high-resolution data for modeling and analysis. For example, the type of commodity being carried must be known in the freight modeling framework. Unfortunately, this information is not available at the state or metropolitan level, or it is expensive to obtain from current resources.

Nevertheless, using …

The Effects Of Mixed-Initiative Visualization Systems On Exploratory Data Analysis, Adam Kern

McKelvey School of Engineering Graduate Student Theses & Dissertations

The main purpose of information visualization is to act as a window between a user and data. Historically, this has been accomplished via a single-agent framework: the only decisionmaker in the relationship between visualization system and analyst is the analyst herself. Yet this framework arose not from first principles, but from necessity: prior to this decade, computers were limited in their decision-making capabilities, especially in the face of large, complex datasets and visualization systems. This thesis aims to present the design and evaluation of a mixed-initiative system that aids the user in handling large, complex datasets and dense visualization systems. …

Paired Trial Classification: A Novel Deep Learning Technique For Mvpa, Jacob M. Williams, Ashok Samal, Prahalada K. Rao, Matthew R, Johnson

School of Computing: Faculty Publications

Many recent developments in machine learning have come from the field of “deep learning,” or the use of advanced neural network architectures and techniques. While these methods have produced state-of-the-art results and dominated research focus in many fields, such as image classification and natural language processing, they have not gained as much ground over standard multivariate pattern analysis (MVPA) techniques in the classification of electroencephalography (EEG) or other human neuroscience datasets. The high dimensionality and large amounts of noise present in EEG data, coupled with the relatively low number of examples (trials) that can be reasonably obtained from a sample …

Event-Based Visual-Inertial Odometry Using Smart Features, Zachary P. Friedel

Theses and Dissertations

Event-based cameras are a novel type of visual sensor that operate under a unique paradigm, providing asynchronous data on the log-level changes in light intensity for individual pixels. This hardware-level approach to change detection allows these cameras to achieve ultra-wide dynamic range and high temporal resolution. Furthermore, the advent of convolutional neural networks (CNNs) has led to state-of-the-art navigation solutions that now rival or even surpass human engineered algorithms. The advantages offered by event cameras and CNNs make them excellent tools for visual odometry (VO). This document presents the implementation of a CNN trained to detect and describe features within …

Retiming Smoke Simulation Using Machine Learning, Samuel Charles Gérard Giraud Carrier

Theses and Dissertations

Art-directability is a crucial aspect of creating aesthetically pleasing visual effects that help tell stories. A particularly common method of art direction is the retiming of a simulation. Unfortunately, the means of retiming an existing simulation sequence which preserves the desired shapes is an ill-defined problem. Naively interpolating values between frames leads to visual artifacts such as choppy frames or jittering intensities. Due to the difficulty in formulating a proper interpolation method we elect to use a machine learning approach to approximate this function. Our model is based on the ODE-net structure and reproduces a set of desired time samples …

Treatment Effects Of Modafinil For Cocaine Use Disorders: A Retrospective Analysis Of Aggregated Clinical Trial Data From Three Cocaine Treatment Studies, Daniel Ruskin

Honors Scholar Theses

Approximately 913,000 individuals in the United States meet the diagnostic criteria for cocaine use disorder (CUD). The widespread usage of cocaine, along with the negative cardiac and neurological effects associated with the drug, has made cocaine one of the top three drugs associated with overdose deaths in the United States. This epidemic has brought cocaine dependency into the public spotlight and has prompted extensive research into treatment strategies. However, at the time of writing, no drugs have been approved by the United States Food and Drug Administration (FDA) for use in treating CUD. The purpose of this study is to …

Graph Classification With Kernels, Embeddings And Convolutional Neural Networks, Monica Golahalli Seenappa, Katerina Potika, Petros Potikas

Faculty Publications, Computer Science

In the graph classification problem, given is a family of graphs and a group of different categories, and we aim to classify all the graphs (of the family) into the given categories. Earlier approaches, such as graph kernels and graph embedding techniques have focused on extracting certain features by processing the entire graph. However, real world graphs are complex and noisy and these traditional approaches are computationally intensive. With the introduction of the deep learning framework, there have been numerous attempts to create more efficient classification approaches. We modify a kernel graph convolutional neural network approach, that extracts subgraphs (patches) …

Using Logical Specifications For Multi-Objective Reinforcement Learning, Kolby Nottingham

Undergraduate Honors Theses

In the multi-objective reinforcement learning (MORL) paradigm, the relative importance of environment objectives is often unknown prior to training, so agents must learn to specialize their behavior to optimize different combinations of environment objectives that are specified post-training. These are typically linear combinations, so the agent is effectively parameterized by a weight vector that describes how to balance competing environment objectives. However, we show that behaviors can be successfully specified and learned by much more expressive non-linear logical specifications. We test our agent in several environments with various objectives and show that it can generalize to many never-before-seen specifications.

Data Mining Of Chinese Social Networks: Factors That Indicate Post Deletion, Meisam Navaki Arefi

Computer Science ETDs

Widespread Chinese social media applications such as Sina Weibo (Chinese Twitter), the most popular social network in China, are widely known for monitoring and deleting posts to conform to Chinese government requirements. Censorship of Chinese social media is a complex process that involves many factors. There are multiple stakeholders and many different interests: economic, political, legal, personal, etc., which means that there is not a single strategy dictated by a single government authority. Moreover, sometimes Chinese social media do not follow the directives of government, out of concern that they are more strictly censoring than their competitors.

One crucial question …

Machine Learning For Effective Parkinson's Disease Diagnosis, Brennon Brimhall

Undergraduate Honors Theses

Parkinson’s Disease is a degenerative neurological condition that affects approximately 10 million people globally. Because there is currently no cure, there is a strong motivation for research into improved and automated diagnostic procedures. Using Random Forests, a computer can effectively learn to diagnose Parkinson’s disease in a patient with high accuracy (94%), precision (95%), and recall (91%) across the data of over 2800 patients. Using similar techniques, I further determine that the most predictive medical tests relate to tremors observed in patients.

Quantitative Analysis Of Evaluation Criteria For Generative Models, Marvin W. Newlin

Theses and Dissertations

Machine Learning (ML) is rapidly becoming integrated in critical aspects of cybersecurity today, particularly in the area of network intrusion/anomaly detection. However, ML techniques require large volumes of data to be effective. The available data is a critical aspect of the ML process for training, classification, and testing purposes. One solution to the problem is to generate synthetic data that is realistic. With the application of ML to this area, one promising application is the use of ML to perform the data generation. With the ability to generate synthetic data comes the need to evaluate the “realness” of the generated …

A Machine Learning Approach To Characterizing Particle Morphology In Nuclear Forensics, Daniel A. Gum

Theses and Dissertations

A machine learning approach is taken to characterizing a group of synthetic uranium bearing particles. SEM images of these lab-created particles were converted into a binary representation that captured morphological features in accordance with a guide established by Los Alamos National Laboratory. Each particle in the dataset contains an association with chemical creation conditions: processing method, precipitation temperature and pH, calcination temperature are most closely tied to particle morphology. Additionally, trained classifiers are able to relate final products between particles, implying that morphological features are shared between particles with similar composition.

Algorithm Selection Framework: A Holistic Approach To The Algorithm Selection Problem, Marc W. Chalé

Theses and Dissertations

A holistic approach to the algorithm selection problem is presented. The “algorithm selection framework" uses a combination of user input and meta-data to streamline the algorithm selection for any data analysis task. The framework removes the conjecture of the common trial and error strategy and generates a preference ranked list of recommended analysis techniques. The framework is performed on nine analysis problems. Each of the recommended analysis techniques are implemented on the corresponding data sets. Algorithm performance is assessed using the primary metric of recall and the secondary metric of run time. In six of the problems, the recall of …

Characterizing Regime-Based Flow Uncertainty, John L. Fioretti

Theses and Dissertations

The goal of this work is to develop a regime-based quantification of horizontal wind field uncertainty utilizing a global ensemble numerical weather prediction model. In this case, the Global Ensemble Forecast System Reforecast (GEFSR) data is utilized. The machine learning algorithm that is employed is the mini-batch K-means clustering algorithm. 850 hPa Horizontal flow fields are clustered and the forecast uncertainty in these flow fields is calculated for different forecast times for regions across the globe. This provides end-users quantified flow-based forecast uncertainty.

Automated Detection And Mitigation Of Inefficient Visual Searching Using Electroencephalography And Machine Learning, Joshua P. Gallaher

Theses and Dissertations

Decisions made during the high-stress and fast-paced operations of the military are extremely prone to cognitive biases. A commonly known cognitive bias is a confirmation bias, or the inappropriate bolstering of an unknown hypothesis. One such critical military operation that can fall prey to a confirmation bias is a visual search. During a visual search, a military operator must perform a visual scan of an environment for a specific target. However, the visual search process can fall prey to the same confirmation bias which can cause inefficient searches. This study elicits inefficient visual search patterns and applies various mitigation techniques …

Identifying Viruses From Metagenomic Data Using Deep Learning, Jie Ren, Kai Song, Chao Deng, Nathan A. Ahlgren, Jed A. Fuhrman, Yi Li, Xiaohui Xie, Ryan Poplin, Fengzhu Sun

Biology

Background: The recent development of metagenomic sequencing makes it possible to massively sequence microbial genomes including viral genomes without the need for laboratory culture. Existing reference-based and gene homology-based methods are not efficient in identifying unknown viruses or short viral sequences from metagenomic data. Methods: Here we developed a reference-free and alignment-free machine learning method, DeepVirFinder, for identifying viral sequences in metagenomic data using deep learning. Results: Trained based on sequences from viral RefSeq discovered before May 2015, and evaluated on those discovered after that date, DeepVirFinder outperformed the state-of-the-art method VirFinder at all contig lengths, achieving AUROC 0.93, 0.95, …

Is It Possible To Discern Striga Weed (Striga Hermonthica) Infestation Levels In Maize Agro-Ecological Systems Using In-Situ Spectroscopy?, Bester Tawona Mudereri, Timothy Dube, Saliou Niassy, Emily Kimathi, Tobias Landmann, Zeyaur Khan, Elfatih M. Abdel-Rahman

All Peer-Reviewed Publications

The invasion by Striga in most cereal crop fields in Africa has posed a significant threat to food security and has caused substantial socioeconomic losses. Hyperspectral remote sensing is an effective means to discriminate plant species, providing possibilities to track such weed invasions and improve precision agriculture. However, essential baseline information using remotely sensed data is missing, specifically for the Striga weed in Africa. In this study, we investigated the spectral uniqueness of Striga compared to other co-occurring maize crops and weeds. We used the in-situ FieldSpec® Handheld 2™ analytical spectral device (ASD), hyperspectral data and their respective narrow-band indices …

Improving The Background Of Gravitational-Wave Searches For Core Collapse Supernovae: A Machine Learning Approach, M. Cavaglià, S. Gaudio, T. Hansen, K. Staats, M. Szczepanczyk, M. Zanolin

Physics Faculty Research & Creative Works

Based on the priorO1-O2observing runs, about30%of the data collected by Advanced LIGO and Virgo Internext observing runs are expected tobe single-interferometer data, i.e. they will be collected at times when only one detector in the network is operating in observing mode. Searches for gravitational-wave signals from supernova events do not rely on matched filtering techniques because of the stochastic nature of the signals. If a Galactic supernova occurs during single-interferometer times, separation of its unmodelled gravitational-wave signal from noise will be even more difficult due to lack of coherence between detectors. We present a novel machine learning method to perform …

Dictionary Learning For Image Reconstruction Via Numerical Non-Convex Optimization Methods, Lewis M. Hicks

University Honors Theses

This thesis explores image dictionary learning via non-convex (difference of convex, DC) programming and its applications to image reconstruction. First, the image reconstruction problem is detailed and solutions are presented. Each such solution requires an image dictionary to be specified directly or to be learned via non-convex programming. The solutions explored are the DCA (DC algorithm) and the boosted DCA. These various forms of dictionary learning are then compared on the basis of both image reconstruction accuracy and number of iterations required to converge.

Artificial Intelligence: A New Paradigm In Obstetrics And Gynecology Research And Clinical Practice, Pulwasha Iftikhar, Marcela V. Kuijpers, Azadeh Khayyat, Aqsa Iftikhar, Maribel Degouvia De Sa

Publications and Research

Artificial intelligence (AI) is growing exponentially in various fields, including medicine. This paper reviews the pertinent aspects of AI in obstetrics and gynecology (OB/GYN) and how these can be applied to improve patient outcomes and reduce the healthcare costs and workload for clinicians.

Herein, we will address current AI uses in OB/GYN, and the use of AI as a tool to interpret fetal heart rate (FHR) and cardiotocography (CTG) to aid in the detection of preterm labor, pregnancy complications, and review discrepancies in its interpretation between clinicians to reduce maternal and infant morbidity and mortality. AI systems can be used …

Machine Learning Predicts Reach-Scale Channel Types From Coarse-Scale Geospatial Data In A Large River Basin, Hervé Guillon, Colin F. Byrne, Belize A. Lane, Samuel Sandoval Solis, Gregory B. Pasternack

Publications

Hydrologic and geomorphic classifications have gained traction in response to the increasing need for basin-wide water resources management. Regardless of the selected classification scheme, an open scientific challenge is how to extend information from limited field sites to classify tens of thousands to millions of channel reaches across a basin. To address this spatial scaling challenge, this study leverages machine learning to predict reach-scale geomorphic channel types using publicly available geospatial data. A bottom-up machine learning approach selects the most accurate and stable model among∼20,000 combinations of 287 coarse geospatial predictors, preprocessing methods, and algorithms in a three-tiered framework to …

Cyber-Physical Security With Rf Fingerprint Classification Through Distance Measure Extensions Of Generalized Relevance Learning Vector Quantization, Trevor J. Bihl, Todd J. Paciencia, Kenneth W. Bauer Jr., Michael A. Temple

Faculty Publications

Radio frequency (RF) fingerprinting extracts fingerprint features from RF signals to protect against masquerade attacks by enabling reliable authentication of communication devices at the “serial number” level. Facilitating the reliable authentication of communication devices are machine learning (ML) algorithms which find meaningful statistical differences between measured data. The Generalized Relevance Learning Vector Quantization-Improved (GRLVQI) classifier is one ML algorithm which has shown efficacy for RF fingerprinting device discrimination. GRLVQI extends the Learning Vector Quantization (LVQ) family of “winner take all” classifiers that develop prototype vectors (PVs) which represent data. In LVQ algorithms, distances are computed between exemplars and PVs, and …

Analysis Of Identification Method For Bacterial Species And Antibiotic Resistance Genes Using Optical Data From Dna Oligomers, Ryan L. Wood, Tanner Jensen, Cindi Wadsworth, Mark Clement, Prashant Nagpal, William G. Pitt

Faculty Publications

Bacterial antibiotic resistance is becoming a significant health threat, and rapid identification of antibiotic-resistant bacteria is essential to save lives and reduce the spread of antibiotic resistance. This paper analyzes the ability of machine learning algorithms (MLAs) to process data from a novel spectroscopic diagnostic device to identify antibiotic-resistant genes and bacterial species by comparison to available bacterial DNA sequences. Simulation results show that the algorithms attain from 92% accuracy (for genes) up to 99% accuracy (for species). This novel approach identifies genes and species by optically reading the percentage of A, C, G, T bases in 1000s of short …

Analysis Of Identification Method For Bacterial Species And Antibiotic Resistance Genes Using Optical Data From Dna Oligomers, Ryan L. Wood, Tanner Jensen, Cindi Wadsworth, Mark Clement, Prashant Nagpal, William G. Pitt

Faculty Publications

Bacterial antibiotic resistance is becoming a significant health threat, and rapid identification of antibiotic-resistant bacteria is essential to save lives and reduce the spread of antibiotic resistance. This paper analyzes the ability of machine learning algorithms (MLAs) to process data from a novel spectroscopic diagnostic device to identify antibiotic-resistant genes and bacterial species by comparison to available bacterial DNA sequences. Simulation results show that the algorithms attain from 92% accuracy (for genes) up to 99% accuracy (for species). This novel approach identifies genes and species by optically reading the percentage of A, C, G, T bases in 1000s of short …