Develop AI systems to efficiently learn new skills and solve open-ended problems, rather than depend exclusively on AI systems trained with extensive datasets
How this post is connected to other assets
Connect the dots with a gray line. Start at red (or green) and go to yellow then the last. Use Manhattan geometry. Make sure to keep the original points as they are.
Hard to think of a physical process where we could simulate this or inform us of the rules.
It's not pathfinding. It's not like a human reasoning about how to complete the line needs to make any choices about where to draw, it's just clear based on some rules and because we can see the whole input at once.
We need the whole input because we need to know the direction of the next point we're going towards, when to stop drawing when we're parallel, then to change directions to get to the point. This can still be calculated using a local (as opposed to a non-local representation) representation.
Discover other posts like this one