At Penn State, I majored in Mathematics, and took a series of courses in discrete mathematics, linear programming, linear algebra, combinatorics, and stochastic modeling. Beyond my math coursework, I also took classes in data science, statistics, and technical writing to round out my education. Below is a selection of projects from my undergraduate coursework that best capture my computational, statistical, and communication abilities.

### Speech Emotion Recognition (PDF)

For an applied data science course, my partner and I were tasked with developing a machine learning model to correctly classify the emotion behind speech clips from the Berlin Emotional Dataset. Here, we developed a novel method for detecting the emotion in a speech clip and compare our method to an ensembled classifier and a convolutional neural network. While our method did not have superior accuracy, it is more robust and scalable than the other models tested. This project awarded the Best Paper Award within the class for the novelty of our method and the clarity of our writing.

### Violent Crimes Regression (PDF)

As a term project for a regression course, I developed a multiple regression model to predict the number of violent crimes in US communities using demographic data from the 1990s supplied by the UCI Machine Learning Repository. To best fit the data, we employed a number of regression techniques, including normalization, weighted least squares, and backward elimination. Our final model was able to explain 66% of the variability in violent crimes per capita, a surprisingly high amount for real socioeconomic data. From this project, I learned the challenges of fitting regression models to real life data.

### Automated Theorem Proving Literature Review (PDF)

In this literature review, I explored the state of machine learning techniques and artificial intelligence concepts in modern automated theorem provers to determine if there is compelling evidence to suggest that these applications of ML/AI are successful. Then, I determined the value behind machine learning and AI inspired automated theorem provers, which tend to be conceptually strong, but the supporting evidence may no longer be relevant. Fundamentally, I have found that there does exist compelling evidence to suggest that machine learning and artificial intelligence can be applied to automated reasoning.