My Awesome CSCI 0451 Blog

Unsupervised Learning: 2 Ways

In this post, we will be exploring unsupervised learning through two examples. We will be working with Singular Value Decomposition (SVD) to do image compression and with Spectral Community Detection to deal with clusters of data. As a result, this blog post is broken into those two sections to explore each topic seperately.

Implementing the Perceptron Algorithm

Engaging with Dr. Timnit Gebru

This week, our machine learning class has the opportunity to hear from Dr. Timnit Gebru, a leader in the field of algorithmic bias. Dr. Gebru is the founder of the Distributed Artificial Intelligence Research Institute (DAIR) and the co-founder of Black in AI. Her work gained national attention after she left Google, where she co-lead the Ethical Artificial Intelligence Team, after the company had issues with a paper she co-authored on the dangers of large language models. She is the recipient of numerous awards and accolades for her work on bias both within technologies and within tech companies. I am very excited and thankful for the opportunity to speak with her in our class setting and to hear her speak to a broader audience at the college on Monday evening.

Optimization for Logistic Regression

In this blog, I will explore optimization for logistic regression through the implementation of three optimization algorithms: simple gradient descent, a momentum method, and stochastic gradient descent. Through experimentation we will compare the performance of each of these algoritms for training logistic regression to predict binary classifiers for a data set.

Auditing Alocative Bias

In this blog post, we are going to examine algorithmic bias through an audit. Using data from the American Community Survey’s Public Use Microdata Sample (PUMS). I will perform an audit on racial bias in a machine learning model that predicts whether or not an individual in employed.To do this, I will begin by downloading data and training a model to make such predictions. Then, we will examine some of the different measures of fairness like predictive parity and error rate before discussing how the model could be biased and what implications that could have in deployment and beyond. This audit will only consider data from New York State.

Linear Regression

Source code: https://github.com/kate-kenny/kate-kenny.github.io/blob/main/posts/blog4/linear_regression.py

Classifying Palmer Penguins

In this blog post, we will explore how different models can be used to classify species of penguins in the Palmer Penguin data set and visualize both our results and some of the decisions that went into the models.

Urban Air Pollution Predicted by Income and Racial Demographics

Final project on determining urban air pollution by income and racial demographics.

Bridget Ulian, Kate Kenny, Mia Tarantola