Joshua Lee — UC San Diego
Email: jdlee@ucsd.edu Section & Mentor: B21 — Albert Hsiao
What is the most interesting topic covered in your domain this quarter?
The most interesting topic covered in my domain this quarter to me is honestly the entire original paper that we are trying to recreate in general. It’s interesting to see how much information you can get from an image by feeding it into a model with a certain level of accuracy.
Describe a potential investigation you would like to pursue for your Quarter 2 Project.
A potential investigation I would like to pursue for my Quarter 2 project would be either trying to implement LLMs into our current project or trying to pursue explainability for our project.
What is a potential change you’d make to the approach taken in your current Quarter 1 Project?
A potential change I’d make to the approach taken in my current Quarter 1 Project would be testing different weights at different resolutions and seeing if that improves the accuracy drop-off at the higher resolutions.
What other techniques would you be interested in using in your project?
Other techniques that I would be interested in using in my project would be testing batch size versus Pearson r as well as figuring out a way to increase the data amount even more.
Site generated with GitHub Pages + Jekyll.