Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I bet you could still train a neural net to complete the levels only using the 1D slices as input.


Would it even be that much more difficult for the net than 2D Mario?


My gut says it'll be a tad more difficult since you won't have any data beyond what's directly in front of mario. Maybe the AI will end up doing a bunch of rapid peeks to determine next move, especially when there are large pits.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: