Burning Silicon

Research Scut Work

Hello World!

I hope you are all doing well! Or at least as well as is possible during COVID. I finally got both vaccines and have waited the two weeks, so I am now immune! And UC Davis has established that fall quarter will be in person! So I'm very much looking forward to physically attending classes because online classes suck. Also I miss hanging out in Kemper Hall! It's kinda depressing, but exactly what you expect a CS building to look like. Because CS people are sad. At least in college.

Machine Learning Research

So I got switched to a different research project! This is because my PI, Professor Tagkopolous, decided that I had done enough just floating around on my own with very little guidance and moved me to a more mature project where I could work closely with a graduate student. At first I was kinda sad because everybody says that it's important to learn to do research on your own. But then I was told by a bunch of grad students that usually you don't get results that way, usually in order to do stuff on your own and get results you need to already be in grad school, unless you are very lucky/talented. So I'm on a new project!

Anyways, despite loving research, there is a part of research that I really do not like: the freaking data analysis. You see, I work with machine learning. And a very basic way to look at machine learning is that you have a system that takes in data, learns from it, and gives predictions on new data — basically like a human. Well, somebody needs to compile that data. Which means that currently, I am going over research papers (I can't go into too much detail because NDA) and taking data out of them. Most research papers are poorly written. They hide essential data in the most hidden locations. So for the past several weeks, my work has been to go over the freaking papers and find the data and put it in a spreadsheet. It SUCKS.

That said, we're about to start doing logistic regression which is cool. Basically, the way it works is, you put in a bunch of data, and you say which inputs resulted in which outputs. It then tries to find which inputs are correlated with which outputs. The problem is, this usually involves a 'black box'. This means that it tells you the answers, but doesn't tell you why you got them.

logistic regression image

So instead of Python, which I'm a big fan of, we gotta use R! Which I don't know how to use! But I'll figure it out, R isn't that hard.

Prepping for CMU

So this summer, as I probably mentioned before, I will be working at CMU! I am somewhat alarmed at this development, but mostly excited. So one way I have been preparing is by trying to get all my work done beforehand. Unfortunately, that does not look like it will be the case, and I will be doing 2 labs at once? I'm gonna die.

D&D figurines

So my housemate who usually runs the D&D game is super busy this quarter and hasn't been running it. I'm kinda sad because I have a really cool Kenku (bird woman) who is very zen and says morbid things all the time but in a happy way. And I want to play her! But since I can't, I decided to buy a some figurines for my characters from this one site Heroforge (I promise they aren't sponsoring me) and I got some cool ones! From left to right they are: Carxes Caryinax, my Triton Paladin, Mavet the Kenku, who is a Death Sorcerer, and Daket Darkbranch, my Kobold Warlock.

dnd figuines
May you be ever victorious in your endeavors.