This project is 3 different machine learning models that attempt to classify paintings by artist. This project was developed as a final project for my first machine learning course, UVM CS 254. In order to make the project complex enough, we decided to work together to create a pipeline for loading in our images/data, and then individually develop our models from there. We all decided to use the transfer learning approach, which involves collecting a previously trained advanced neural net, setting its layers to not be trainable, then adding a couple more layers after it, and training those layers on our dataset. This proved to be quite effective. For a more descriptive write-up on our project, visit the GitHub repository, linked below.
One of the more exciting aspects on this project was being able to use the Vermont Advanced Computing Core. When working with image data, especially a lot of it and large models, training the model on our own computers quickly proved to be infeasible. As such, we were granted access to the VACC by our professor, which we utilized to train our models. This taught me how to use a batch computer, an experience I had not had before. For more information about the VACC, visit https://www.uvm.edu/vacc.
For a more descriptive write-up on the project, visit the GitHub page for it, linked below.