Let’s run a tiny chess neural network by hand! A few years ago someone tried to learn chess in a month and then play world champion Magnus Carlsen. He had the wild idea of hand running a neural network on a piece of paper during the game! This probably isn’t physically possible (…he didn’t win), but what’s the closest we could get? This made me curious: what’s the smallest neural network you could make that could be helpful for a human? What if you only had pen and paper (and maybe a lot of time), or if you had a calculator? We’ll go over the smallest possible neural nets for chess. In the era of mega NNs these will be refreshingly tiny! Amédée is a full stack software engineer from Montréal (now moving to Amsterdam!). They like growing culinary mushrooms, playing chess, and techno music. Presented at !!Con 2024 (pronounced, of course, "bangbangcon"!) in Santa Cruz, CA -- https://bangbangcon.com/ !!Con is a production of the Exclamation Foundation -- https://exclamation.foundation/ Video Production by Confreaks Follow Confreaks 👇 https://confreaks.com https://x.com/confreaks
Get notified about new features and conference additions.