Physical Sciences and Mathematics Commons^™

An Introduction To The Analysis Of Ranked Response Data, Holmes Finch

Practical Assessment, Research, and Evaluation

Researchers in many disciplines work with ranking data. This data type is unique in that it is often deterministic in nature (the ranks of items k-1 determine the rank of item k), and the difference in a pair of rank scores separated by k units is equivalent regardless of the actual values of the two ranks in the pair. Given its unique qualities, there are specific statistical analyses and models designed for use with ranking data. The purpose of this manuscript is to demonstrate a strategy for analyzing ranking data from sample description through the modeling of relative ranks ...

Power Properties Of Ordinal Regression Models For Likert Type Data, Ulf Olsson

Practical Assessment, Research, and Evaluation

We discuss analysis of 5-grade Likert type data in the two-sample case. Analysis using two-sample t tests, nonparametric Wilcoxon tests, and ordinal regression methods, are compared using simulated data based on an ordinal regression paradigm. One thousand pairs of samples of size n=10 and n=30 were generated, with three different degrees of skewness. For all sample sizes and degrees of skewness, the ordinal probit model has highest power. This is not surprising since the data was generated with this model in mind. Slightly more surprising is that the t test has higher power than the Wilcoxon test in ...

Utilizing Climate Change Refugia For Climate Change Adaptation And Management In The Northeast, Sara A. Wisner

Masters Theses

To account for the effects of climate change, management plans in the northeast need to incorporate climate adaptation. Conserving climate change refugia is one adaptation strategy. Climate change refugia are areas buffered by climate change that enable the persistence of valued physical, ecological, and cultural resources; preserving these areas could be a potential adaptation strategy. Using a translational ecology approach where researchers and managers from the National Park Service, US Geological Survey, the University of Massachusetts, and elsewhere worked together, we focused on identifying refugia for tree, herbaceous plant, mammal, and bird species in order to prioritize them for conservation ...

Metareasoning For Planning And Execution In Autonomous Systems, Justin Svegliato

Doctoral Dissertations

Metareasoning is the process by which an autonomous system optimizes, specifically monitors and controls, its own planning and execution processes in order to operate more effectively in its environment. As autonomous systems rapidly grow in sophistication and autonomy, the need for metareasoning has become critical for efficient and reliable operation in noisy, stochastic, unstructured domains for long periods of time. This is due to the uncertainty over the limitations of their reasoning capabilities and the range of their potential circumstances. However, despite considerable progress in metareasoning as a whole over the last thirty years, work on metareasoning for planning relies ...

Anticanonical Models Of Smoothings Of Cyclic Quotient Singularities, Arie A. Stern Gonzalez

Doctoral Dissertations

In this thesis we study anticanonical models of smoothings of cyclic quotient singularities. Given a surface cyclic quotient singularity $Q\in Y$, it is an open problem to determine all smoothings of $Y$ that admit an anticanonical model and to compute it. In \cite{HTU}, Hacking, Tevelev and Urz\'ua studied certain irreducible components of the versal deformation space of $Y$, and within these components, they found one parameter smoothings $\Y \to \A^1$ that admit an anticanonical model and proved that they have canonical singularities. Moreover, they compute explicitly the anticanonical models that have terminal singularities using Mori's ...

Reliable Decision-Making With Imprecise Models, Sandhya Saisubramanian

Doctoral Dissertations

The rapid growth in the deployment of autonomous systems across various sectors has generated considerable interest in how these systems can operate reliably in large, stochastic, and unstructured environments. Despite recent advances in artificial intelligence and machine learning, it is challenging to assure that autonomous systems will operate reliably in the open world. One of the causes of unreliable behavior is the impreciseness of the model used for decision-making. Due to the practical challenges in data collection and precise model specification, autonomous systems often operate based on models that do not represent all the details in the environment. Even if ...

Modeling Chain Packing In Complex Phases Of Self-Assembled Block Copolymers, Anugu Abhiram Reddy

Doctoral Dissertations

Block copolymer (BCP) melts undergo microphase seperation and form ordered soft matter crystals with varying domain shapes and symmetries. We study the con- nection between diblock copolymer molecular designs and thermodynamic selection of ordered crystals by modeling features of variable sub-domain geometry filled with individual blocks within non-canonical sphere-like and network phases that together with layered, cylindrical and canonical spherical phases forms “natural forms” of self- assembled amphiphilic soft matter at large. First, we present a model to revise our understanding of optimal Frank-Kasper sphere-like morphologies by advancing the- ory to account for varying domain volumes. We then develop generic ...

Decision Making With Limited Data, Kieu My Phan

Doctoral Dissertations

This thesis studies different approaches to decision making with limited data.

First, we study the effects of approximate inference on Thompson sampling in the k-armed bandit problems. Thompson sampling is a successful algorithm but requires posterior inference, which often must be approximated in practice. We show that even small constant inference error (in alpha-divergence) can lead to poor performance (linear regret) due to under-exploration (for alpha < 1) or over-exploration (for alpha > 0) by the approximation. While for alpha > 0 this is unavoidable, for alpha <= 0 the regret can be improved by adding a small amount of forced exploration.

Second, we consider the problem of designing a randomized experiment on a source population to estimate the Average Treatment Effect (ATE ...

Moving Polygon Methods For Incompressible Fluid Dynamics, Chris Chartrand

Doctoral Dissertations

Hybrid particle-mesh numerical approaches are proposed to solve incompressible fluid flows. The methods discussed in this work consist of a collection of particles each wrapped in their own polygon mesh cell, which then move through the domain as the flow evolves. Variables such as pressure, velocity, mass, and momentum are located either on the mesh or on the particles themselves, depending on the specific algorithm described, and each will be shown to have its own advantages and disadvantages. This work explores what is required to obtain local conservation of mass, momentum, and convergence for the velocity and pressure in a ...

Mixture Models In Machine Learning, Soumyabrata Pal

Doctoral Dissertations

Modeling with mixtures is a powerful method in the statistical toolkit that can be used for representing the presence of sub-populations within an overall population. In many applications ranging from financial models to genetics, a mixture model is used to fit the data. The primary difficulty in learning mixture models is that the observed data set does not identify the sub-population to which an individual observation belongs. Despite being studied for more than a century, the theoretical guarantees of mixture models remain unknown for several important settings.

In this thesis, we look at three groups of problems. The first part ...

Decision-Analytic Models Using Reinforcement Learning To Inform Dynamic Sequential Decisions In Public Policy, Seyedeh Nazanin Khatami

Doctoral Dissertations

We developed decision-analytic models specifically suited for long-term sequential decision-making in the context of large-scale dynamic stochastic systems, focusing on public policy investment decisions. We found that while machine learning and artificial intelligence algorithms provide the most suitable frameworks for such analyses, multiple challenges arise in its successful adaptation. We address three specific challenges in two public sectors, public health and climate policy, through the following three essays.

In Essay I, we developed a reinforcement learning (RL) model to identify optimal sequence of testing and retention-in-care interventions to inform the national strategic plan “Ending the HIV Epidemic in the US ...

Incremental Non-Greedy Clustering At Scale, Nicholas Monath

Doctoral Dissertations

Clustering is the task of organizing data into meaningful groups. Modern clustering applications such as entity resolution put several demands on clustering algorithms: (1) scalability to massive numbers of points as well as clusters, (2) incremental additions of data, (3) support for any user-specified similarity functions.

Hierarchical clusterings are often desired as they represent multiple alternative flat clusterings (e.g., at different granularity levels). These tree-structured clusterings provide for both fine-grained clusters as well as uncertainty in the presence of newly arriving data. Previous work on hierarchical clustering does not fully address all three of the aforementioned desiderata. Work on ...

Few-Shot Natural Language Processing By Meta-Learning Without Labeled Data, Trapit Bansal

Doctoral Dissertations

Humans show a remarkable capability to accurately solve a wide range of problems efficiently -- utilizing a limited amount of computation and experience. Deep learning models, by stark contrast, can be trained to be highly accurate on a narrow task while being highly inefficient in terms of the amount of compute and data required to reach that accuracy. Within natural language processing (NLP), recent breakthroughs in unsupervised pretraining have enabled reusable models that can be applied to many NLP tasks, however, learning of new tasks is still inefficient. This has led to research on few-shot learning, where the goal is to ...

The Thermoelectric, Thermoresistive, And Hygroresistive Properties And Applications Of Vapor Printed Pedot-Cl, Linden K. Allison

Doctoral Dissertations

Wearable electronics are a valuable tool to increase consumer access to real-time and long-term health care monitoring. The development of these technologies can also lead to major advancements in the field, such as self-charging systems that are completely removed from the electrical grid. However, much of the wearable technology available commercially contain rigid components, use unsustainable synthetic methods, or undesirable materials. The field has thus been moving towards wearables that mimic textiles or use textiles as a substrate. Herein, we discuss the use of oxidative chemical vapor deposition (oCVD) to produce textiles coated with poly(3,4-ethylenedioxythiophene) known as PEDOT-Cl ...

Improving The Hydrological Analysis Of Groundwater Flow Paths By Integrating Geochemical And Physical Characteristics Of A Highly Fractured Aquifer System To Create Sustainable Use Of Groundwater In A Climate With Projected Drying Trends., Marsha K. Allen

Doctoral Dissertations

Improving the hydrological analysis of groundwater flow paths by integrating geochemical and physical characteristics of a highly fractured aquifer system to create sustainable use of groundwater in a climate with projected drying trends.

Precipitation over Caribbean islands has decreased steadily since the 1950's, which has led to severe drought conditions. The most recent Pan-Caribbean Drought occurred from 2013 to 2016. Climate models predict that drying trends are expected to continue and become more severe over time as precipitation decreases and temperatures rise. In addition, evaporation rates on these islands are expected to increase by ~15-17%, contributing to the drought ...

Assessment Of The Economic And Ecosystem Service Contributions Of Usda Forest Service Landowner Assistance Programs In The Conterminous United States, Jacqueline S. Dias

Masters Theses

Forests provide immense goods and services to both local and regional communities. The USDA Forest Service’s, State and Private Forestry program administer multiple landowner assistance programs aimed at helping private forest owners while supporting the continued supply of ecosystem services (e.g., timber harvesting, recreation, carbon sequestration and storage). The two landowner assistance programs assessed in this study are the Forest Legacy Program (FLP) and the Forest Stewardship Program (FSP). A majority of the nation’s forests are privately owned, many of which are facing deleterious impacts like wildfires, invasive species, development pressures, and other adverse effects from climate ...

Breaking Down Barriers To Consistent, Climate-Smart Regulation Of Invasive Plants - A Case Study Of Northeast States, Bethany A. Bradley, Evelyn M. Beaury, Emily Fusco, Lara Munro, Carrie Brown-Lima, William Coville, Benjamin Kesler, Nancy Olmstead, Jocelyn Parker

Environmental Conservation Faculty Publication Series

Efforts to prevent the introduction and spread of new invasive plants are most effective when regulated species are consistent across jurisdictional boundaries and proactively prohibit species before they arrive or in the earliest stages of invasion. Consistent and proactive regulation is particularly important in the northeast U.S. which is susceptible to many new invasive plants due to climate change. Unfortunately, recent analyses of state regulated plant lists show that regulated species are neither consistent nor proactive. To understand why, we focus on two steps leading to invasive plant regulation across six northeast states (Connecticut, Maine, Massachusetts, New Hampshire, New ...

Correlational Analysis Of Mammals And Residential Land Use: Amherst, Ma, Ella Gutkowski

Massachusetts GIS Day

Mammal diversity varies in different types of land uses. Residential land use oftentimes interferes with the natural occurrence of mammal species. This study conducts a correlational analysis using camera trap data from Excel and land use data in GIS to uncover whether humans have disrupted mammal occurrence in residential land use areas in Amherst, MA. Results reveal that human activity in residential land use areas in Amherst did not strongly influence the occurrence of these mammals.

Understanding The Comparative Fit Index: It's All About The Base!, Saskia Van Laar, Johan Braeken

Practical Assessment, Research, and Evaluation

Despite the sensitivity of fit indices to various model and data characteristics in structural equation modeling, these fit indices are used in a rigid binary fashion as a mere rule of thumb threshold value in a search for model adequacy. Here, we address the behavior and interpretation of the popular Comparative Fit Index (CFI) by stressing that its metric for model assessment is the amount of misspecification in a baseline model and by further decomposition into its fundamental components: sample size, number of variables and the degree of multivariate dependence in the data. Simulation results show how these components influence ...

Georeferencing The Macconnell Aerial Photo Collection, Alex Heilmann, Matthew Martin, Camille Barchers, Forrest J. Bowlick, Rebecca M. Seifried

Massachusetts GIS Day

In the 1950s, Professor William P. MacConnell from the University of Massachusetts Forestry Department began working with his students to map the land cover in Massachusetts via the state’s earliest aerial photography program. These individual photographs are now part of the Special Collections and University Archives at the University of Massachusetts Amherst Libraries, and although they have been digitized and made available online, they have not yet been georeferenced.

In Spring 2021, our team (Alex and Matthew) began manually georeferencing the photos in ArcMap 10.8 software onto USGS 2019 color orthoimagery of Massachusetts available from MassGIS. Ideal ground ...

Data For "Relic Groundwater And Mega Drought Confound Interpretations Of Water Sustainability And Lithium Extraction In Arid Lands", Brendan J. Moran, David F. Boutt, Sarah V. Mcknight, Jordan Jenckes, Lee Ann Munk, Daniel Corkran, Alexander Kirshen

Data and Datasets

This repository contains raw data from this publication including hydrogeochemistry, model calculations, groundwater levels, and remotely sensed data compiled and extracted using Google Earth Engine.

Understanding Of Visual Domains Via The Lens Of Natural Language, Chenyun Wu

Doctoral Dissertations

A joint understanding of vision and language can enable intelligent systems to perceive, act, and communicate with humans for a wide range of applications. For example, they can assist a human to navigate in an environment, edit the content of an image through natural language commands, or search through image collections using natural language queries. In this thesis, we aim to improve our understanding of visual domains through the lens of natural language. We specifically look into (1) images of categories within a fine-grained taxonomy such as species of birds or variants of aircraft, (2) images of textures that describe ...

Modeling And Characterization Of Optical Metasurfaces, Mahsa Torfeh

Masters Theses

Metasurfaces are arrays of subwavelength meta-atoms that shape waves in a compact and planar form factor. During recent years, metasurfaces have gained a lot of attention due to their compact form factor, easy integration with other devices, multi functionality and straightforward fabrication using conventional CMOS techniques. To provide and evaluate an efficient metasurface, an optimized design, high resolution fabrication and accurate measurement is required. Analysis and design of metasurfaces require accurate methods for modeling their interactions with waves. Conventional modeling techniques assume that metasurfaces are locally periodic structures excited by plane waves, restricting their applicability to gradually varying metasurfaces that ...

Human Mobility Monitoring Using Wifi: Analysis, Modeling, And Applications, Amee Trivedi

Doctoral Dissertations

Understanding and modeling humans and device mobility has fundamental importance in mobile computing, with implications ranging from network design and location-aware technologies to urban infrastructure planning. Today's users carry a plethora of devices such as smartphones, laptops, tablets, and smartwatches, with each device offering a different set of services resulting in different usage and mobility leading to the research question of understanding and modeling multiple user device trajectories. Additionally, prior research on mobility focuses on outdoor mobility when it is known that users spend 80% of their time indoors resulting in wide gaps in knowledge in the area of ...

Windows In Algebraic Geometry And Applications To Moduli, Sebastian Torres

Doctoral Dissertations

We apply the theory of windows, as developed by Halpern-Leistner and by Ballard, Favero and Katzarkov, to study certain moduli spaces and their derived categories. Using quantization and other techniques we show that stable GIT quotients of $(\mathbb{P}^1)^n$ by $PGL_2$ over an algebraically closed field of characteristic zero satisfy a rare property called Bott vanishing, which states that $\Omega^j_Y \otimes L$ has no higher cohomology for every j and every ample line bundle L. Similar techniques are used to reprove the well known fact that toric varieties satisfy Bott vanishing. We also use windows to explore ...

Learning From Limited Labeled Data For Visual Recognition, Jong-Chyi Su

Doctoral Dissertations

Recent advances in computer vision are in part due to the widespread use of deep neural networks. However, training deep networks require enormous amounts of labeled data which can be a bottleneck. In this thesis, we propose several approaches to mitigate this in the context of modern deep networks and computer vision tasks.

While transfer learning is an effective strategy for natural image tasks where large labeled datasets such as ImageNet are available, it is less effective for distant domains such as medical images and 3D shapes. Chapter 2 focuses on transfer learning from natural image representations to other modalities ...

United States Household Carbon Footprints: Quantifying The Relationship Between Household-Level Income Inequality And Greenhouse Gas Emissions (1996-2015), Jared Starr

Doctoral Dissertations

As long as humanity has existed, we have altered our environment to provide goods, services, and (more recently) wealth to people. Over the last several centuries, the scope and pace of this transformation has accelerated with the onset of technological innovation, social and economic reorganization, and an ensuing population boom. Today, humanity’s demands on nature have become the dominant force shaping the critical earth systems upon which all life depends. From local land-use change to the global climate many of these anthropogenic pressures pose an existential threat to nature and the dependent social systems that rely on them. Yet ...

Mathematical Model For Osteosarcoma Progression And Treatments, Trang M. Le

Doctoral Dissertations

Cancer is a complex disease where every tumor has its own characteristics, and thus different tumors may respond differently to the same treatments. Osteosarcoma, which is a rare type of cancer with poor prognosis, is especially characterized by its high heteogeneity. Therefore, it is important to study the progression of osteosarcoma tumors in different groups of patients with distinct characteristics. The immune system has been reported to play an important role in the development of various cancers with some immune cells having anti-tumor effects and others having pro-tumor effects. With recent advances in digital cytometry methods, which are techniques to ...

Equivariant Smoothings Of Cusp Singularities, Angelica Simonetti

Doctoral Dissertations

Let $p \in X$ be the germ of a cusp singularity and let $\iota$ be an antisymplectic involution, that is an involution free on $X\setminus \{p\}$ and such that there exists a nowhere vanishing holomorphic 2-form $\Omega$ on $X\setminus \{p\}$ for which $\iota^*(\Omega)=-\Omega$. We prove that a sufficient condiition for such a singularity equipped with an antisymplectic involution to be equivariantly smoothable is the existence of a Looijenga (or anticanonical) pair $(Y,D)$ that admits an involution free on $Y\setminus D$ and that reverses the orientation of $D$.

Deep Learning Models For Irregularly Sampled And Incomplete Time Series, Satya Narayan Shukla

Doctoral Dissertations

Irregularly sampled time series data arise naturally in many application domains including biology, ecology, climate science, astronomy, geology, finance, and health. Such data present fundamental challenges to many classical models from machine learning and statistics. The first challenge with modeling such data is the presence of variable time gaps between the observation time points. The second challenge is that the dimensionality of the inputs can be different for different data cases. This occurs naturally due to the fact that different data cases are likely to include different numbers of observations. The third challenge is that different irregularly sampled instances have ...