Week of 1/30
Good morning my friends! Or afternoon, I guess. This week we got started with our projects.
In the beginning of last week, we went to CalTech and met with Professor Hassibi's two graduate students. We pitched some of our ideas to them-- from our e-cigarette sensor to Yelp recommender system to our Sign Language Translator, each suggestion was rebuked with some very grounded insight. We eventually decided that with only a few months remaining in the semester, we would focus on a more realistic project to pursue. Based on the graduate student's recommendations, we decided to work on a supervised and unsupervised version of a game called Dots and Boxes. The supervised version would be our realistic project and the unsupervised version would be our ambitious project.
We chose to recreate Dots and Boxes because of its ability to keep people entertained for a very long time. The objective of the game is to draw as many boxes as possible and outnumber your opponents in the number of boxes you can create. Initially, we wanted to do Tic Tac Toe. However, Tic Tac Toe was already a solved game, and it wouldn't make sense to integrate understandings of machine learning or reinforcement learning to it. Dots and Boxes is a nuanced enough game that people have made a bot to play it but not with machine learning, as per my research. Therefore, I think that this'll be an educational endeavor for the entire team to take on.
In the beginning of last week, we went to CalTech and met with Professor Hassibi's two graduate students. We pitched some of our ideas to them-- from our e-cigarette sensor to Yelp recommender system to our Sign Language Translator, each suggestion was rebuked with some very grounded insight. We eventually decided that with only a few months remaining in the semester, we would focus on a more realistic project to pursue. Based on the graduate student's recommendations, we decided to work on a supervised and unsupervised version of a game called Dots and Boxes. The supervised version would be our realistic project and the unsupervised version would be our ambitious project.
Dots and Boxes!
We chose to recreate Dots and Boxes because of its ability to keep people entertained for a very long time. The objective of the game is to draw as many boxes as possible and outnumber your opponents in the number of boxes you can create. Initially, we wanted to do Tic Tac Toe. However, Tic Tac Toe was already a solved game, and it wouldn't make sense to integrate understandings of machine learning or reinforcement learning to it. Dots and Boxes is a nuanced enough game that people have made a bot to play it but not with machine learning, as per my research. Therefore, I think that this'll be an educational endeavor for the entire team to take on.
Reinforcement Learning is an Enigma!
This helped me understand reinforcement learning a lot better! :)
I'm working on learning about reinforcement learning right now in order to apply it to the code that my group is working on, but I think I might had jumped too far into the deep end in my initial research. A lot of math symbols came up that I had no idea what it said. I think I should tune it down a notch and try to understand the big picture of reinforcement learning before I try to investigate specific algorithms and equations I can use for this project. Despite the difficulty of it all, I'm determined to actually learn about reinforcement learning and how it actually works. It seems really cool-- and it'll be magnificent if I can eventually decipher the code!
Looking at the Bellman equation is already giving me a headache! Yikes!
On a completely unrelated note, I've recently found myself really overloaded with stress. I think I might have bitten off more than I can chew in for my course scheduling-- and now that second semester is starting, I'm really starting to feel the edge that people often describe in agony about during their Junior year. After a series of disastrous missteps (and getting the flu), I think that my best option right now is to take things one step at a time and to just concentrate and work my [butt] off this semester. I've never found myself more productive than now, and while that's a good thing, I feel so incredibly worn out. Drawing is a hobby that I often turn to when I'm feeling stressed, and I want to show you some of the work I've done (thirty minutes before I wrote this):
I think I might just call this one "Violet Melancholia"
Maybe this'll be "in my mind"
It's a pool!
Comments
Post a Comment