3 ms·
grandmasters typically play a small number of (recorded) games per year, would it be possible to train a neural network on lots of games to recognize what were
by bryanrasmussen 6y ago
grandmasters typically play a small number of (recorded) games per year, would it be possible to train a neural network on lots of games to recognize what were likely moves of a player from a small number of games - so that you could have a computer to work with that would prepare you for a match with a grandmaster?
- gameswithgo 6y agolately grandmasters play a huge number of recorded games per year online
- wildlogic 6y agoyeah, these days you can tune into a twitch stream and grab some top grandmaster games any day of the week
- bryanrasmussen 6y agohuh, even better, although guess I'm behind the times.
- akamoonknight 6y agoI'm wondering something similar, where maybe GMs could train against a neural network that is built off of their upcoming opponent's historical games and thus they would get more experience against that 'opponent'. Instead of having to actually have played against the opponent themselves to learn weaknesses.
- mimimi31 6y agoI think there is an app that claims to let you play against Magnus Carlsen at different ages. I always assumed this is how they imemented that feature.