applied_machine_learning_public

Project 2

T-shirt Design:

The stylized image seen on this t-shirt was created by combining the style of a Jackson Pollock piece with the content of an image of a brain using a total variation loss method over 15 epochs.

I chose the image of the brain for two reasons. One, data science is compromised of thinking logically about complex problems. This starts with the brain, although the computer allows us to perform this type of thinking on a much larger scale. In fact, this course itself is about teaching machines how to learn like a brain in order to make predictions and analyze data about the world around us. Two, this data we analyze is often about ourselves and our brains: our effect on the environment, the social sciences, etc. Data science is about converting our collective brain power into a larger, mechanized process in order to fully discover the world around us and how it works.

I used the Jackson Pollock painting Convergence as my stylized image, as I believe that fitting models often feels like throwing paint at a wall and then taking a step back to see if it stuck. There are so many adjustments that need to be made to create the best model, and finding the correct adjustments often starts with just trying things: adding more layers, preprocessing the data, changing the number of neurons in each layer, checking for overfitting, and much more. To me, this process mirrors that of Jackson Pollock. Pollock is known for is paint-splattering technique. He chooses a color, splatters it willy-nilly, and then examines the work he’s created and what needs adjusting. Just as Pollock’s method is an art form, so is data science. Furthermore, I specifically chose one of Pollock’s more colorful paintings, as data science is not as black and white as it may seem, as there are ethical concerns and gray areas throughout the discipline.