Timothy Martin Nicdao

For this project, the task was to evaluate the readability of text and rate them based on their difficulty. The target values that I tried to predict were the Bradley-Terry analysis of pairwise comparisons between excerpts. Teachers spanning grades 3-12 served as the raters for these comparisons with majority being teachers for grades 6-10.

In order to evaluate the readability of the texts, I used a hugging face Transformer Roberta model to extract features of the text and added a regression head for the down stream task. The model was implemented with PyTorch and to determine model generalizability, I used a Kfold cross validation technique with 5 folds.

This project was built for the class ICS 432: Concurrency and was implemented in Java. Our task was to build an image processing app that uses concurrency to speed up processing of batches of images. For this project, we had to implement several methods that allow for concurrency without allowing race conditions by using Locks, Condition Variables and Semaphores.

Are we alone in the Universe? The SETI Breakthrough Listen - E.T. Signal Search aims to answer this question by applying machine learning algorithms to make sense of telescope data.
The data consists of signals of telescopes in a form of 2 dimensional arrays and the objective is to classify positive signals that were artificially created in order to build systems that can discern patterns in the signals.
For this project, I tried using convolutional neural networks with a 5 fold cross validation.