Every story and comment from the Hacker News API, in a layout borrowed from 2001.
A comment on AI Index 2021. Parent comment
If you scramble all pixels in some game in a random, but fixed way, you will never learn to play it.
But speed of training of a simple feed forward network will not change at all.
Here you have somewhat simpler version of what I propose: https://high-level-4.herokuapp.com/experiment
Here is the paper: https://rach0012.github.io/humanRL_website/
There are no comments on this comment yet.
If you scramble all pixels in some game in a random, but fixed way, you will never learn to play it.
But speed of training of a simple feed forward network will not change at all.
Here you have somewhat simpler version of what I propose: https://high-level-4.herokuapp.com/experiment
Here is the paper: https://rach0012.github.io/humanRL_website/