3 ms·
On my path to improve myself and learn more about machine learning and where it could be applied, I came up with this idea: to feed the neural network during ex
by atum47 8y ago
On my path to improve myself and learn more about machine learning and where it could be applied, I came up with this idea: to feed the neural network during execution time.
On the firsts round, it plays randomly. After 3 rounds gathering information about the player it starts making predictions of next move.
- cyborgx7 8y agoVery interesting. I once learned some basic statistics about how most people play rock-paper-scissors, and can now pretty reliably win most best of 3 games I play. I had to unlearn all of that strategy for this, instead trying to guesstimate the statistical model of my playstyle it had developed and subvert it. The computer and I actually stayed pretty close together in terms of points, eventhough the computer was slightly ahead of me. The number of ties was only a third of the other 2 point values, which tells me it isn't just randomness.
- atum47 8y agoIf you start with a pattern and then change later on the game, it might take a while for the nn to catch up, but it'll eventually.