Physical Sciences and Mathematics Commons^™

Exploring Loss Functions In Machine Learning, Yujie Wang

CGU Theses & Dissertations

The loss function plays a critical role in machine learning. It is fundamental in training, evaluating, and optimizing machine learning models, directly impacting their effectiveness and efficiency in solving specific tasks. We explore three new loss functions and their applications. Softmax Cross-Entropy Loss, stands as a prevalent choice in neural network classification tasks. It treats all misclassifications uniformly. However, multi-class classification problems often have many semantically similar classes. We should expect that these semantically similar classes will have similar parameter vectors. We introduce a weighted loss function, the tree loss as a drop-in replacement for the cross entropy loss. The …

Fr1: Comics, Cyborgs, And “In Between” Identities, Ella Lehavi

Scripps Senior Theses

As a queer Jew who grew up surrounded by immigrant cultures and communities, I find myself in a liminal space between my identities and the dominant culture of my country– one where my perspective on gender and my cultural experiences aren’t fully understood by the world I exist in. Comics and cartoons are an explorational platform for concepts of reality and identity; they are one of very few spaces where I see my identities explored with so much depth and care.

Cartoons and comics exist in between realistic depictions and abstraction. This makes them a great place to express all …

Transition Towards A Brighter Future: Energy Usage Analysis At Pilgrim Place And Recommendations For Decarbonization, Zhuoyao Wu

Scripps Senior Theses

This thesis summarizes the semester-long project I and other Environmental Analysis major students collaborated on for the EA 190 Capstone Project aims to find effective methods to reduce emissions and increase energy efficiency. Our group partnered with the Climate Resilience Initiative team (CRI) at Pilgrim Place, a social and environmental justice-focused retirement community in Claremont, CA. As the demand for effective methods for energy efficiency in communal spaces increases, we have conducted an energy analysis and researched potential functional technologies to replace high fossil fuel consumption sources. Our final deliverable is a written report that analyzes the data we collected …

Probing The Ising Model’S Thermodynamics Through Restricted Boltzmann Machines, Xiaobei (Emma) Zhang

HMC Senior Theses

This thesis explores the connection between physics and machine learning by using Restricted Boltzmann Machines (RBMs) to study the thermodynamic properties of the Ising model. The Ising model is a simple but realistic model that captures the magnetic behavior of a system, where spins occupy a lattice of sites and different spin configurations correspond to different energies. The model exhibits phase transitions between ferromagnetic and paramagnetic phases as a function of temperature. RBMs are two-layered neural networks that can learn probability distributions over binary spins. The study generates 2D Ising model data at different temperatures using Monte Carlo simulations, including …

Exploring Sigmoidal Bounded Confidence Models With Mean Field Methods, Tian Dong

HMC Senior Theses

Mathematicians use models of opinion dynamics to describe how opinions in a group of people change over time, which can yield insight into mechanisms behind phenomena like polarization and consensus. In these models, mathematicians represent the community as a graph, where nodes represent agents and edges represent possible interactions. Opinion updates are modeled with a system of differential equations (ODEs). Our work focuses on the sigmoidal bounded confidence model (SBCM), where agents update their opinion toward a weighted average of their neighbors' opinions by weighting similar opinions more heavily. Using tools developed in physics (mean-field theory), we derive a continuity …

Decreasing Cross-Coupling In The Tianlai Radio Antenna Array, Nadia Marie Schwartz Bolef

Scripps Senior Theses

The 21-cm line is the electromagnetic radiation spectral line created by the change in energy state of neutral hydrogen atoms. This signal can be used for intensity mapping, a method of mapping the large-scale structure of the universe at different redshifts. However, challenges are being faced in the designing and building of the radio antenna arrays used to detect the 21-cm line signal. Currently, cross-coupling between antenna feeds is introducing beam errors and correlated noise that could overpower the signal. One possible solution we are working on is to use a phased array of antenna feeds instead of dishes and …

Improving Accessibility And Understanding: Studying The Impact Of Human Activities And Dam Removals On Anadromous Fish Species In The Pacific Northwest, Adia F. Bennett

Scripps Senior Theses

Human activities in North American watersheds have significantly impacted anadromous fish, causing some species of salmon and steelhead to become threatened and endangered. These fish hold ecological, cultural, and economic importance, yet factors such as habitat destruction, overfishing, pollution, climate change, and hydroelectric dams continue to degrade populations. Dams are particularly harmful to anadromous fish, sparking a recent movement advocating for dam removals. This study uses NOAA’s 2022 Biological Viability Assessment to examine the response of anadromous fish to recent dam removals in the Pacific Northwest. Four populations were selected to analyze due to their interaction with sites that had …

Geographically Dependent Thermodynamic And Geochemical Evolution In The Post-Batholith Upper Mantle Beneath The Sierra Nevada: Analysis Of The Blue Knob Xenoliths, Eleanor Ruos

Scripps Senior Theses

Xenoliths entrained in recent (Miocene) volcanics from the Sierra Nevada provide insights into a continuously evolving upper mantle during post-batholith generation periods (early Cenozoic–present). Hypothesized mantle refertilization followed by subsequent delamination of the lithospheric root beneath the Sierra provides a contextual framework for analyzing how the upper mantle responds to such events. This study contributes new geochemical data from the ultramafic xenoliths of Blue Knob, including bulk-rock major and trace element data, thin-section petrography, mineral chemistry and thermobarometry, and oxygen isotopes (delta¹⁸O). Blue knob (Eastern Sierra) data are compared to previous works on xenoliths from Chinese Peak and …

C. Compactum Acts As A Comprehensive Climate Archive And Ecological Foundation In The Labrador Sea, Sadie Heckman

CMC Senior Theses

Clathromorphum compactum, a species of crustose coralline algae (CCA), is incredibly valuable for the future of high latitude ocean health, both as a comprehensive archive of changing ocean conditions, and ecologically as a foundational species for promoting biodiversity. Previous work establishes C. compactum as an effective climate proxy, and its life history provides several advantages for this use. C. compactum grow in nongeniculate, generally radial formations on hard substrates, over a wide distribution in mid-to-high latitude oceans and at subtidal depth ranges. Indeterminate growth leads to extreme longevity in C. compactum (Halfar et al., 2008), and growth rates are relatively …

Towards Algorithmic Justice: Human Centered Approaches To Artificial Intelligence Design To Support Fairness And Mitigate Bias In The Financial Services Sector, Jihyun Kim

CMC Senior Theses

Artificial Intelligence (AI) has positively transformed the Financial services sector but also introduced AI biases against protected groups, amplifying existing prejudices against marginalized communities. The financial decisions made by biased algorithms could cause life-changing ramifications in applications such as lending and credit scoring. Human Centered AI (HCAI) is an emerging concept where AI systems seek to augment, not replace human abilities while preserving human control to ensure transparency, equity and privacy. The evolving field of HCAI shares a common ground with and can be enhanced by the Human Centered Design principles in that they both put humans, the user, at …

The Effects Of Seaweed Farming On Local Ecosystems In The United States, Reid Cohen

Pitzer Senior Theses

In my thesis, I explore how seaweed farming contributes to the regeneration of marine ecosystems, supports biodiversity, and assists in harmful nutrient removal. Seaweed farming is a relatively unexplored industry in the United States with huge carbon capturing potential. Seaweed farming poses a significant opportunity for a project that has the potential to sequester carbon, increase biodiversity, protect coastal ecosystems, naturally clean water, provide biofuel capabilities, substitute as livestock fodder which then decreases methane emissions, along with many more benefits. Kelp supports biodiversity in marine ecosystems and provides essential habitats for various species. The seaweed farming industry holds profound potential …

Drone-Based Topographic Monitoring Of The Doheny Beach Replenishment Project As An Alternative To Land-Based Monitoring, Miller Mccraw

CMC Senior Theses

The rising threat of coastal erosion to California’s beach ecosystems and economy has fueled a rise in coastal stabilization projects, including beach replenishment. This process’s potentially adverse impact on a beach’s topography and ecosystem makes post-replenishment monitoring essential for long-term coastline management. Drone-based monitoring presents itself as a faster, cheaper, and safer alternative to traditional post-replenishment monitoring but has little proof of concept as a practical substitute. This study used drone-based photogrammetry coupled with publicly available wave data to track elevation changes at Doheny and San Capistrano Beach after a beach replenishment project to both determine the beach’s resilience to …

Exploring U.S. Natural Disasters And Psychological Distress: From Time Series Trends To Machine Learning Insights On Hurricane Helene, Sarah Jane Fullerton

CMC Senior Theses

This research investigates the historical trends of psychological distress in the U.S. in relation to natural disaster occurrences. By analyzing long-term data, we examine how significant natural disasters relate to levels of psychological distress over time. The research employs Exploratory Data Analysis (EDA) and Time Series Analysis to identify patterns and trends between the frequency and intensity of natural disasters and the rise of psychological distress across various periods in U.S. history. Additionally, real-time data from Reddit was collected through a custom-built Reddit web scraper specialized for Hurricane Helene. This dataset was labeled for sentiment and used to train machine …

Reducing Generalization Error In Multiclass Classification Through Factorized Cross Entropy Loss, Oleksandr Horban

CMC Senior Theses

This paper introduces Factorized Cross Entropy Loss, a novel approach to multiclass classification which modifies the standard cross entropy loss by decomposing its weight matrix W into two smaller matrices, U and V, where UV is a low rank approximation of W. Factorized Cross Entropy Loss reduces generalization error from the conventional O( sqrt(k / n) ) to O( sqrt(r / n) ), where k is the number of classes, n is the sample size, and r is the reduced inner dimension of U and V.

Synergetic Effects Of Democracy And Economic Development On Income Inequality And The Role Of Relative Political Capacity, Nazif Sali

CGU Theses & Dissertations

Diving into the complex dynamics of income inequality, this dissertation studies the multifaceted relationships between democracy, economic development, and inequality, placing a spotlight on the mediating influence of Relative Political Capacity (RPC). Uncovering the pivotal role of RPC as democracies advance and economies progress, this research navigates the contours of income inequality. By dissecting distinct sub-samples of OECD and non-OECD countries, nuanced insights surface. Employing robust methodologies such as ordinary least squares regressions and fixed effects analysis, I argue that targeted policies addressing inequality can not only foster inclusive economic growth but also fortify the foundations of democratic institutions. As …

Applications And Analysis Of Nlp Deep Learning Models For Antibiotic's Side Effects Classifications, An Ly

CGU Theses & Dissertations

I introduce a novel recursive modification to the classical Goemans-Williamson MaxCut algorithm, offering improved performance in vectorized data clustering tasks. Focusing on the clustering of medical publications, I suggest to employ recursive iterations in conjunction with a dimension relaxation method to enhance density of clustering results. Furthermore, I propose a new vectorization technique for articles, leveraging conditional probabilities for more effective clustering. I believe that these methods will provide advantages in both computational efficiency and clustering accuracy. I will analyze the effectiveness of recursive iterations and higher-dimensional generalizations of the GWA in the hopes of achieving more accurate dissimilarity-based clustering. …

Existence Of Well-Defined Pointer Observable Selects Tensor Product Factorizations Of Quantum Systems, Brian Lee

CMC Senior Theses

In the decoherence account of quantum mechanics, a choice of particular tensor product structure (a particular partition of system into subsystems) is assumed. We explore whether it is possible to relax this arbitrary choice by requiring that a valid tensor product structure admits a quasi-classical description. Such tensor product structures are said to be quasi-classical or decoherence-selected tensor product structures. This project generalizes a 2-qubit quasi-classical tensor product structure selection algorithm to an n-qubit selection algorithm, which allows us to, for the first time, consider the relationship between decoherence-selected tensor product structures and locality-selected tensor product structures. To generalize the …

Unveiling The Power Of Shor's Algorithm: Cryptography In A Post Quantum World, Dylan Phares

CMC Senior Theses

Shor's Algorithm is an extremely powerful tool, in utilizing this tool it is important to understand how it works and why it works. As well as the vast implications it could have for cryptography

Climate And Cognitive Performance: Examining The Relationship Between Extreme Temperatures And Students’ Academic Achievement In The United States, Sophia Kestell

CMC Senior Theses

This study explores how exposure to high temperatures throughout the school year affects students' academic performance in the United States, focusing on third- and eighth-graders and economically disadvantaged students. This paper uses math and reading assessment data for students from the Standard Education Data Archive (SEDA) and temperature data from the PRISM Climate Group at Oregon State. Using a fixed-effects ordinary least squares (OLS) regression model specification, this analysis investigates the relationship between temperature threshold and academic achievement in 333 US counties from 2010 through 2019. Additionally, county air conditioning coverage data are used to explore the relationship between temperatures …

Determining The 2022-2023 Mass Balance Of The Palisade Glacier In The Sierra Nevada Mountains Of California With Remote Sensing, Density Modeling, And Temperature-Index Techniques, Vijay Jain, Eric Grosfils, Sarah Kavassalis

CMC Senior Theses

Small, alpine glaciers, such as those in the Sierra Nevada, are difficult to study because of their small size and remoteness, however, they are important recorders of the impacts of climate change in temperate, alpine environments. Previous studies have attempted to characterize the health of these glaciers using extent change techniques, but these methodologies can only roughly approximate the rudimentary measurement of changing ice volume. This thesis uses the Airborne Snow Observatory Inc.’s (ASO’s) aerial lidar snow depth datasets to perform a mass balance calculation for the Palisade Glacier over the remarkable 2022-2023 water year (October 1 through September 30), …

The Future Of Hydropower In Southeast Asia’S Renewable Energy Transition, Olivia Cairns

CMC Senior Theses

This thesis will examine the barriers to implementing hydropower in Southeast Asia and propose responsible solutions. This is necessary because Southeast Asia is not on track to reduce carbon emissions to meet climate change mitigation goals. Coal often outcompetes renewable energy sources as a means of addressing the growing demand for electricity. Many insist that the potential for solar and wind power expansion is high, but often ignore the key problems with these renewables at an industrial scale. Hydroelectric dams face rampant social opposition, mistrust due to policy failures, and a poor reputation for causing social and environmental harm. Yet, …

The Impact Of The Use Of Mercury In Mining In Colombia, Johan Martinez

CMC Senior Theses

Artisanal and small-scale gold mining is a practice that is common all around the world, especially in Latin America, Africa, and Asia. It is also one of the leading consumers of mercury and leading sources of emissions. Governments have banned the use of mercury completely, including Colombia, but it continues to be used and emitted into the environment. This issue is exacerbated because there is a lack of standards, a lack of research done regarding mercury use in mining, and a complex socio-political history in Colombia. For this reason, this paper seeks to provide a proposed pathway for research to …

Improveing F-Beta Score In Classifying Shark Data Into Shark Behaviors, Ibrahim M. Ali

CGU Theses & Dissertations

One metric used to measure classification performance in machine learning is F-beta score. The objective in this thesis is to improve the average F-b score computed in classifying shark data into shark behaviors, namely; Resting, Swimming, Feeding, and Non-Directed Motion (NDM). Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic Sampling (ADASYN) are utilized to balance the data, from which pre-processed Fast Fourier Transform (FFT), Walsh-Hadamard Transform (WHT), and Autocorrelation (AC) features are extracted then classified using Convolutional Neural Network (CNN) and K-Nearest Neighbors (K-NN). All the combinations of the two balancing techniques, the three feature types, and the two machine …

Reducing Short-Chain Pfas Levels In California Water Supplies, Manu Prabandham

Pomona Senior Theses

This thesis proposes twelve specific policies based on precedents set by prior regulation of persistent organic pollutants (such as PCBs), the costs and benefits of short-chain PFAS, technologies available to remove and destroy short-chain PFAS, and the roles and limitations of California’s regulatory institutions. The twelve policies are chosen to be politically and financially feasible, effective at removing short-chain PFAS from water supplies, and equitable towards lower-income and minority Californians, who suffer the most from the consequences of PFAS and other environmental pollutants. The new definitions, education campaigns, studies, taxes, bans, standards, testing, and filtration systems proposed are intended to …

Mechanisms Of Dendritic Shorting In Lithium Metal Batteries With Li7la3zr2o12 Solid Electrolytes, Lukas Karapin-Springorum

Pomona Senior Theses

Energy storage will play a crucial role in efforts to mitigate the effects of climate change caused by greenhouse gas emissions from human activity. Lithium metal batteries using solid electrolytes like Li₇La₃Zr₂O₁₂ have higher energy density than lithium-ion batteries, which may enable a more rapid and complete electrification of transportation. However, lithium metal batteries suffer from undesirable short-circuiting when metallic lithium deposits connect the two electrodes. It has been debated whether these lithium dendrites generally grow directionally from the anode or are generated by the reduction of lithium ions inside the solid electrolyte, …

Generative Adversarial Networks For Music Generation, Harry Berman

Pomona Senior Theses

In this paper, we aim to harness a machine learning model called Genera- tive Adversarial Networks (GAN) in order to produce AI generated musical strands. The “Generative” part of the model’s name implies that its goal is to create something – music in the case of this paper – and the “Adversarial Networks” refer to the fact that there are two neural networks that learn from each other. One network attempts to trick the other one by creating music that it believes sounds real while the other tries to discern the real from the fake music. After training for long …

Investigating Factors Contributing To Differences In Air Pollutants Between Metropolitan Areas In The Us, Ellen Hu

Scripps Senior Theses

Densely populated regions around the United States exhibit varying concentrations of air pollutants that can impact the health of urban populations and surrounding ecosystems. Understanding why these differences exist can be important in addressing health issues as the human population continues to increase and communities across the globe move to accommodate these trends. To investigate the impact of natural processes and anthropogenic sources on pollutant concentrations, EPA daily fine particulate matter (PM_2.5), nitrogen dioxide (NO₂), and ozone (O₃) pollutant concentrations from seven metropolitan regions across the US for 2018-2022 were used. A Spearman Rank-Order …

Nitrate Contamination In Two California Groundwater Basins: What Creates Disparities?, Catherine Protiva

Scripps Senior Theses

Nitrate contamination is a major problem within California groundwater basins. A form of nitrogen, nitrates leach into groundwater aquifers from the surface where they remain for many decades. The most common source for the contaminant is agricultural sources like nitrogen fertilizer. Consequently, the problem is concentrated in regions of the state with a history of large-scale agricultural production. When water is pulled from contaminated basins and used as drinking water, the nitrates can cause serious negative health consequences, most commonly methemoglobinemia, also known as blue baby syndrome. In this thesis, I will examine what factors contribute to a region’s ability …

Bayesian Inference In Reinforcement Learning Neural Networks During A Markov Decision Processes?, Katherine Graham

Scripps Senior Theses

The predictive mind theory proposes that brains work in a way that makes predictions about future stimuli to process information efficiently and accurately. Bayesian brain theory suggests that the brain utilizes Bayesian probability models to make predictions, while the free-energy minimization hypothesis proposes that these predictions are made to minimize energy or uncertainty, ensuring accurate perceptions. Vertechi et al. (2020) explored animal participants’ utilization of stimulus-bound strategy versus inference-based strategy to solve a Markov decision process with a 2-state environment, one of which is always active. These sites have a certain probability of switching to a different site and the …

A Petrographic And Geochemical Study Of The Rangataua Lava Flows, Mount Ruapehu, Aotearoa New Zealand, Lucy Smith

Scripps Senior Theses

Mount Ruapehu is a large stratovolcano on the North Island of Aotearoa New Zealand. The Rangataua lava flow is situated on the southern flanks of Ruapehu, and extends for 15-16 km. The flow is of andesite composition and comprised of three units: proximal, medial and distal. Mount Ruapehu is both scientifically and culturally significant.

Geochemical analyses of 13 whole rock samples, Electron Microprobe analyses of three polished thin sections, and geothermobarometry calculations were conducted to obtain data on pre-eruptive intensive parameters. Petrographic observations were used to investigate disequilibrium textures and shed light on magma chamber processes.

Results of this study …