Computer program that learns to play classic NES games

Channel:suckerpinch · Length:16:17 · Published:2013-04-02 · Views:2209592

Watch on YouTube

Summary

Tom7 explains and demos software that learns to play NES games by finding score patterns in memory and then searching for inputs that raise it. It is the video companion to his SIGBOVIK paper.

Why HN recommends it

Loved on HN as a hilarious, brilliant example of hacker-academic work and cited whenever game-playing AI or fuzzing comes up.

Watch the whole thing, it's delightful. — HN, 2021

Machine learning & deep learning · Game history & retro gaming · Chess, board games & puzzles

Subtitles (excerpt)

Hey, what's up? I'm Tom 7 and this is a video version of my paper for Sigovic 2013. The paper's called The First Level of Super Mario Brothers is easy with lexographic orderings and time travel. After that, it gets a little bit hard. That's also part of the title of the paper. It's a long title. I want to make sure I get everything in there so it's accurate because unlike most Sigovic work, this is a real paper with real results. What this paper is basically about is um you know the Nintendo Entertainment System is a very tiny computer. Uh Nintendo only has 2 kilob of RAM. Tiny tiny amount of RAM. You could see it. Maybe I'll stick something in the video there that's like small so you can see how small bytes is. is this tiny. So I thought what kinds of really simple sort of amusingly simple um techniques can be used to automate Nintendo game playing. What I do, here's the basic idea. I take a um take a video of me playing the game. And by video, I mean I just record the button presses that I uh that I pressed while I was playing the game. So let's say it's Super Mario Brothers. Play Super Mario Brothers a bit. I'd be at like a level. I'm like jumping around Mario, stomping some glmbbas, riding

On Hacker News

There was an AI created the absolutely brilliant Tom7 on on his YouTube channel 'SuckerPinch' on April 1, 2013 that paused a game of Tetris instead of losing. In my opinion, Tom doesn't get enough credit for his work. He released his AI a full 8+ months before DeepMind released their Atari playing AI. Watch the whole thing, it's delightful. The AI pauses Tetris right before it loses at around 16m… — fartcannon, 2021-02
Thanks for pointing that out. I wouldn't have noticed who that was otherwise. His video on learnfun/playfun is both hilarious and amazing. [VIDEO] — learc83, 2018-02
Not quite what you're describing, but no one has yet linked the classic Tom7 series where he applies deep learning to classic NES games: [VIDEO] — rdlw, 2025-02
tom7 is awesome. He made an awesome and hilarious video years ago about his SIGBOVIK paper/research on AI playing NES: [VIDEO] — dan_quixote, 2022-10
I'm not sure many people can pull this off. The author here was also great in his presentation about automated NES playing, [VIDEO] . You don't see often academics & hacker work that is also combined with the right amount of humor. He even 3Dified the original Zelda! — onli, 2018-06

Related videos