Today, I was invited to give a guest lecture for the class CS224N: Natural Language Processing with Deep Learning. I was excited. First, I’d never given a lecture to such a big audience before – there are 400+ students in the class. Second, it’s Richard Socher‘s class. He’s hands down one of the most chill professors I know. For some reason, he always looks like he’s just got out of bed and we occasionally catch him biking down the stairs to the classroom. Third, I’d always heard that speaking at NVIDIA Auditorium is lit and I want to try it out before graduating. Continue reading “[Day 626] I gave a lecture to 400 students”
If you don’t already know, style transfer is the cool, hip thing that has been taking the recreational AI community by storm. It’s so cool that even Kristen Stewart co-authored a paper about it. To quote one researcher who has done extensive work in style transfer that I’ve got a chance to talk to, “it is an utterly unremarkable paper that wouldn’t have been published otherwise [if Kristen Stewart’s name is not on it]. That’s a publicity stunt.”
Some background on why I’m doing this: I’m teaching the course CS 20SI: “TensorFlow for Deep Learning Research” and for the assignment about convolution neural networks, I thought it’d be fun for students to do style transfer as their exercise at home. They, after all, showed a lot of enthusiasm when we did Deep Dream in class.
Last summer, I worked under Richard Stebbing and he is kinda a genius. I googled him the other day and found out that he finished his undergrad engineering degree in 3 years with straight A-plus. He then became a Rhodes scholar at Oxford, finishing his PhD also in 3 years. Every time I see him code, I’m like: “Wow, you can do that?” You can check out his GitHub. But whenever he wasn’t blowing me away with his coding skills, he made a sport out of making fun of me.
For our CS224D’s final project, Lucio and I took on Kaggle’s Automated Essay Scoring competition. We tried to build a model that can automatically grade your essay. You input an essay and voila, it outputs the score for it. The dataset we have is for essays grade 7 to 10, but the model is easily scalable. It can be used to grade SAT/ACT practice essays or any kind of essays, as long as we have enough training data.
When I got my first Mac 5 years ago, my programmer friends almost disowned me for being such a disgrace to the local nerd community. At that time, there was a prevailing sentiment that real coders used Windows or Linux. Macs were for the fuzzy, the uninitiated, the sparkling nincompoop in the realm of marketing.
When I first studied graphics programming, I was traumatized that the coordinate system on a computer’s screen is upside down. The positive x-axis starts on the far left and points to the right as normal coordinates should do, but the y-axis has its 0 at the top of the screen and nosedives straight down to hell from there. Imagine that you have all your graphics worked out beautifully on paper, and then when you try to program it in a computer, you have to flip all the figures and re-calculate all the coordinates. Why can’t computer scientists be normal for once and respect the centuries-old Math? Cartesian coordinates were invented in the 17th century, while the first electronic general-purpose computer (ENIAC) didn’t come out until 1946.
I learned about Evil Hangman a while ago when I was reading my professor’s blog (Don’t judge me. I’m sure you googled your professors too. The Onion wrote about it.). I never got around to write this game until today when I woke up and decided that I should do something with my life.
Evil Hangman is like normal Hangman — players try to guess a secret word by entering different letters. But in Evil Hangman, players are (almost) guaranteed to lose. I said “almost” because Evil Hangman is a program with a deterministic algorithm. If a player knows how it works, they can guess the letters in a way to maximize their chance of winning.