Physical Sciences and Mathematics Commons^™

Understanding Model Reasoning In Automated Speech Systems: Implementing A Prototype Explanation System Using The Lime Method, Vadim Kudlay

Honors Theses

The field of voice processing has seen great advancements thanks in part to the rise of deep learning. However, the application of these deep learning techniques with an audio input space leads to an interesting result not commonly found when dealing with other input domains. Namely, common techniques for generating auditory adversarial samples using gradient-based optimization have been observed to have extremely low transferability among even the same model structure. This implies an inherent difference in the latent representations of audio samples that may be worth investigating in the pursuit of a more resilient and interpretable voice processing framework. Our …

Fact-Checking Of Claims From The English Wikipedia Using Evidence In The Wild, Aalok Sathe

Honors Theses

Automated fact checking is a task in the domain of Natural Language Processing that deals with the verification of claims using evidence. Fact checking is becoming increasingly important as large amounts of human-generated information accumulate online. In the recent past, our society has witnessed large-scale spread of disinformation via the internet that has time and again led to noticeable disruptions in the fabric of society. Fact-checking would help mitigate the spread of disinformation by allowing large magnitudes of content to be automatically evaluated for disinformation. In this work, we construe and tackle multiple subtasks of fact checking using labeled data …

Biasing Medial Axis Rapidly-Exploring Random Trees With Safe Hyperspheres, David Qin

Honors Theses

Motion planning is a challenging and widely researched problem in robotics. Motion planning algorithms aim to not only nd unobstructed paths, but also to construct paths with certain qualities, such as maximally avoiding obstacles to improve path safety. One such solution is a Rapidly-Exploring Random Tree (RRT) variant called Medial Axis RRT that generates the safest possible paths, but does so slowly. This paper introduces a RRT variant called Medial Axis Ball RRT (MABallRRT) that uses the concept of clearance -- a robot's distance from its nearest obstacle -- to efficiently construct a roadmap with safe paths. The safety of …

Fast Medial Axis Sampling For Use In Motion Planning, Hanglin Zhou

Honors Theses

Motion planning is a difficult but important problem in robotics. Research has tended toward approximations and randomized algorithms, like sampling-based planning. Probabilistic RoadMaps (PRMs) are one common sampling-based planning approach, but they lack safety guarantees. One main approach, Medial Axis PRM (MAPRM) addressed this deficiency by generating robot configurations as far away from the obstacles as possible, but it introduced an extensive computational burden. We present two techniques, Medial Axis Bridge and Medial Axis Spherical Step, to reduce the computational cost of sampling in MAPRM and additionally propose recycling previously computed clearance information to reduce the cost of connection in …

Connectiveity And Structures Of Coloring Graphs, Xin Yutong

Honors Theses

Reconfiguration problems have been studied and applied to solve problems in various areas, including Math, Computer Science, and Chemistry. Due to the close relatedness between coloring graphs and reconfiguration problems, the connectivity and structure of coloring graphs give valuable information to a solution set of the corresponding reconfiguration problem. In this paper we will discuss 2--connectedness and cut-vertices of coloring graphs, and forbidden structures on a coloring graph wit cut-vertices.

Computer-Assisted Coloring-Graph Generation And Structural Analysis, Wesley Su

Honors Theses

Graphs are a well studied construction in discrete math, with one of the most common areas of study being graph coloring. The graph coloring problem asks for a color to be assigned to each vertex in a graph such that no two adjacent vertices share a color. An assignment of k colors that meets these criteria is called a k-coloring. The coloring graph Ck(G) is defined as the graph where every vertex represents a valid k-coloring of graph G and edges exist between colorings that di↵er by one vertex. We call graph G the base graph of the k-coloring graph …

Differential Equations Models Of Pathogen-Induced Single- And Multi-Organ Tissue Damage, Fiona Lynch

Honors Theses

The rise of antibiotic resistance has created a significant burden on healthcare systems around the world. Antibiotic resistance arises from the increased use of antibiotic drugs and antimicrobial agents, which kill susceptible bacterial strains, but have little effect on strains that have a mutation allowing them to survive antibiotic treatment, defined as “resistant” strains. With no non-resistant bacteria to compete for resources, the resistant bacteria thrives in this environment, continuing to reproduce and infect the host with an infection that does not respond to traditional antibiotic treatment.

A number of strategies have been proposed to tackle the problem of antibiotic …

Toward A Scientific Investigation Of Convolutional Neural Networks, Anh Tran

Honors Theses

This thesis does not assume the reader is familiar with artificial neural networks. However, to keep the thesis concise, it assumes the reader is familiar with the standard Machine Learning concepts of training set, validation set, and test set [1]. Their usage is intended to help ensure that the Machine Learning system can generalize its training from input examples used during its training to “similar” kinds of examples never used during its training.

The concept of a Convolutional Neural Network (CNN) is one of the most successful computational concepts today for solving image classification problems. However, CNNs are difficult and …

Differential Privacy For Growing Databases, Gi Heung (Robin) Kim

Honors Theses

Differential privacy [DMNS06] is a strong definition of database privacy that provides indi- viduals in a database with the guarantee that any particular person’s information has very little effect on the output of any analysis of the overall database. In order for this type of analysis to be practical, it must simultaneously preserve privacy and utility, where utility refers to how well the analysis describes the contents of the database.

An analyst may additionally wish to evaluate how a database’s composition changes over time. Consider a company, for example, that accumulates data from a growing base of customers. This company …

Nonexistence Of Nonquadratic Kerdock Sets In Six Variables, John Clikeman

Honors Theses

Kerdock sets are maximally sized sets of boolean functions such that the sum of any two functions in the set is bent. This paper modifies the methodology of a paper by Phelps (2015) to the problem of finding Kerdock sets in six variables containing non-quadratic elements. Using a computer search, we demonstrate that no Kerdock sets exist containing non-quadratic six- variable bent functions, and that the largest bent set containing such functions has size 8.

Cameron-Liebler Line Classes And Partial Difference Sets, Uthaipon Tantipongipat

Honors Theses

The work consists of three parts. The first is a study of Cameron-Liebler line classes which receive much attention recently. We studied a new construction of infinite family of Cameron-Liebler line classes presented in the paper by Tao Feng, Koji Momihara, and Qing Xiang (rst introduced in 2014), and summarized our attempts to generalize this construction to discover any new Cameron-Liebler line classes or partial difference sets (PDSs) resulting from the Cameron-Liebler line classes. The second is our approach to finding PDS in non-elementary abelian groups. Our attempt eventually led to the same general construction of PDS presented in John …

Real-Time Translation Of American Sign Language Using Wearable Technology, Jackson Taylor

Honors Theses

The goal of this work is to implement a real-time system using wearable technology for translating American Sign Language (ASL) gestures into audible form. This system could be used to facilitate conversations between individuals who do and do not communicate using ASL. We use as our source of input the Myo armband, an affordable commercially-available wearable technology equipped with on-board accelerometer, gyroscope, and electromyography sensors. We investigate the performance of two different classification algorithms in this context: linear discriminant analysis and k-Nearest Neighbors (k-NN) using various distance metrics. Using the k-NN classifier and windowed dynamic time …

Statistical Analysis Of The Variability And Reliability Of Eye-Tracking Test In Measuring Mild Traumatic Brain Injury, Xi He

Honors Theses

Saccadic eye-tracking tests have been advocated as a useful tool to distinguish mTBI patients from healthy people. However, intra-individual variances sometimes interfere with the interpretation of eye-tracking results, especially in experiments when group size is restricted. This study analyzes eye-tracking results of 14 mTBI patients taking the test twice with no medical administration in between. Using more accurate models to fit each individual's result, variables such as asymptote (of the fit func­tions) and hypothetical values for peak velocity, peak acceleration, and duration are derived for variability analysis. We conclude that the asymptotes for peak velocity and peak acceleration are the …

Detecting Malicious Javascript, Matthew F. Der

Honors Theses

The increased use of the World Wide Web and JavaScript as a scripting language for Web pages have made JavaScript a popular attack vector for infecting users' machines with malware. Additionally, attackers often obfuscate their code to avoid detection, which heightens the challenge and complexity of automated defense systems. We present two analyses of malicious scripts and suggest how they could be extended into intrusion detection systems. For our analyses we use a sample of deobfuscated malicious and benign scripts collected from actual Web sites. First, using our malicious sample, we perform a manual analysis of attack signatures, identifying four …

Boundary Behavior Of Laplace Transforms, Timothy Ferguson

Honors Theses

In this thesis, we examine the boundary behavior of Laplace transforms (as analytic functions on the right and left half planes) of certain bounded functions. The types of bounded functions we consider are Fourier transforms of measures and almost periodic functions.

On Conway's Generalization Of The 3x + 1 Problem, Robin M. Givens

Honors Theses

This thesis considers a variation of the 3x+1, or Collatz, Problem involving a function we call the Conway function. The Conway function is defined by letting C3(n)=2k for n=3k and C3(n)=4k±1 for n=3k±1, where n is an integer. The iterates of this function generate a few 'short' cycles, but the s' tructural dynamics are otherwise unknown. We investigate properties of the Conway function and other related functions. We also discuss the possibility of using the Conway function to generate keys for cryptographic use based on a fast, efficient binary implemenation of the function. Questions related to the conjectured tree-like structure …

Relative Difference Sets In 2-Groups : A Group Cohomological Viewpoint, Brian Wyman

Honors Theses

No abstract provided.

Ecological Niching In An Interactive Simulation, Ryan T. Webb

Honors Theses

Our goal is to create a simulation platform for the study of ecological niching that can be extended to suit the needs of biological research. Ecological niching and the accompanying evolutionary process of speciation are difficult to observe in situ, which makes them prime candidates for study via the methods of computer simulation. To this end, we have created an interactive, real-time ecosystem simulation based on the standard predator/prey interaction model, in which interacting populations of organisms exhibit swarming behavior. We hope to provide the basic simulation components necessary to bring about niching and speciation, that may be extended for …

Factorization Of Polynomials And Real Analytic Function, Radoslaw L. Stefanski

Honors Theses

In this project, we address the question: When can a polynomial p(x, y) of two variables be factored as p(x, y) = f(x)g(y), where f and g are polynomials of one variable. We answer this question, using linear algebra, and create a Mathematica program which carries out this factorization. For example,

3+3x-5x^3+y+xy-5/3x^3y+y^2+xy^2-5/3x^3y^2 = (1+x-5/3x^3)(3+y+y^2)

We then generalize this concept and ask: When can p(x,y) can be written as

p(x,y) = f1(x)g2(y)+f2(x)g2(y)+...+fr(x)gr(y)

where fj,gj are polynomials. This can certainly be done (for large enough r). Which is the minimum such r? Again, we have a Mathematica program which carries out this …

Discovering The Potential For Advancements In Intrusion Detection Systems, Kenneth J. Buonforte

Honors Theses

An intrusion detection system (IDS) is a collection of monitors strategically placed on a network or individual host in order to detect anomalous behavior. Since James Anderson introduced one of the first frameworks for an intrusion detection system in [1]. researchers have extended the capabilities of these systems. IDSs take many forms, from systems as basic as command line audit logs to those that emulate the defense mechanisms of the human immune system. However, as intrusion detection has evolved, scientists and administrators alike are beginning to question the need for this technology due to its lack of quantifiable performance and …

Securing Distributed Computations : In Search Of Reliable Large-Scale Compute Power And Refreshed Redundancy, Edward P. Kenney

Honors Theses

The Internet may be the single largest technological advance or significant societal change in the last century. Not only does it allow access to more information than any human could ever hope to digest, but it produces the potential of having millions of computers combining their computational forces for the betterment of a single cause .. Tl is is the fundamental goal of distributed computing. A distributed system is defined to be a network of machines with some degree of centralized direction. In a distributed computational system each machine will accept computational tasks from a supervisor in a master-slave relationship. …

Self-Similarity In Network Traffic, Francisco Chinchilla

Honors Theses

It is critical to properly understand the nature of network traffic in order to effectively design models describing network behavior. These models are usually used to simulate network traffic, which in turn are used to construct congestion control techniques, perform capacity planning studies, and/or evaluate the behavior of new protocols. Using the wrong models could lead to potentially serious problems such as delayed packet transmissions or an increase in packet drop rates.

Traditionally, packet arrivals were assumed to follow a Poisson arrival process. Although Poisson processes have several properties that make them easy to work with, they do not accurately …

Shout With The Largest Mob : Toward A Model For Primitive Communication In Mobile Automata, Rebecca A. Weber

Honors Theses

We consider the problem of simulating communication between independent, autonomous agents, or machines, using only local rules with no global control over the agents' behavior. First, we construct an algorithm by which the machines will avoid collisions with each other and with boundaries or obstacles. Noting that collision avoidance alone will not result in higher-level behavior, and with the goal of creating agents which would self-organize, we begin to develop a signalling system by which agents can communicate. This leads to a new method for modeling agent motion in the plane. Throughout, we are motivated by possible linkages between our …

Method For Identification Of Origins Of Replication And Genes Regulated By Dnaa In Bacteria, Olga G. Troyanskaya

Honors Theses

The study is focused on developing computer programs to identify origin of DNA replication based on analysis of total bacterial genomes, scoring regions for number of DnaA binding sites, AT content, DNA adenine methylase boxes, and integration host factors binding sites. The programs were tested on cyanobacterium Synechocystis, and several potential origins were identified. However, no one definite region could be located. Currently, software is being developed to analyze common motifs around the origins of all bacteria with known origins. Genes whose transcription could be regulated by DnaA were identified by searching for DnaA boxes preceding promoter regions.

The Dirichlet Problem And Its Physical Motivations, Andrew E. Pitts

Honors Theses

In this work, we explore the basics of harmonic function theory and its relationship to problems in the theory of heat diffusion. In particular, we will focus on the classical Dirichlet problem.

Problems In Harmonic Function Theory, Ronald A. Walker

Honors Theses

Harmonic Function Theory is a field of differential mathematics that has both many theoretical constructs and physical connections, as well as its store of classical problems.

One such problem is the Dirichlet Problem. While the proof of the existence of a solution is well-founded on basic theory, and general methods for polynomial solutions have been well studied, much ground is still yet to be overturned. In this paper we focus on the examination, properties and computation methods and limitations, of solutions for rational boundary functions.

Another area that we shall study is the properties and generalizations of the zero sets …

On Some New Constructions Of Difference Sets, Sarah Agnes Spence

Honors Theses

Difference sets are mathematical structures which arise in algebra and combinatorics, with applications in coding theory. The fundamental question is when and how one can construct difference sets. This largely expository paper looks at standard construction methods and describes recent findings that resulted in new families of difference sets. This paper provides explicit examples of difference sets that arise from the recent constructions. By gaining a thorough understanding of these new techniques, it may be possible to generalize the results to find additional new families of difference sets. The paper also introduces partial and relative difference sets and discusses how …

On The Automatic Generation Of Network Protocol Simulators, Andrew Chen

Honors Theses

Computers communicate with each other over various communication networks via a language known as a protocol. The design of the protocol can have a significant impact on the efficiency (and effectiveness) of the network. Because building an actual network to test the performance (and reliability) of a new protocol is rather expensive and time consuming, there is an interest in simulating network protocols in order to determine how efficient the communication network is. We are therefore interested in automatically generating simulators that could measure the performance of the new protocols. There are two main parts to this project. The first …

Temporal Flocking And Cacophony Simulating Agent Communication In A Noisy Environment, Jessica R. Crawford

Honors Theses

Realistic communication is one of the most difficult aspects of simulating group behavior because the patterns produced by group communication are complex and not easily definable. In this paper, we present a model, developed using artificial life methodology, for creating simulations of group communication. Our model employs autonomous, artificial agents to produce emergent group behavior that resembles the communication patterns of a group, specifically, a flock of birds. Each agent collects information about its environment and its neighbors and follows a set of rules designed to meet both group goals and individual agent goals. Because we seek to establish emergent …

Parallel Programming, Peter Dailey

Honors Theses

The speed of technology is always increasing, especially in the field of computing. Unfortunately, the size of the problems needing to be solved are also growing in many areas. In order to keep up with this, parallel computing has become an important research area. The term parallel computing essentially refers to using multiple processors cooperating to solve a problem. For certain problems this can speed up the solution by a factor ofN, the number of processors being used. There are algorithms, for which there is no speed increase due to certain dependencies.