Certainly if the bot had a representative sample of actions in the kitchen. I think it could digest recipe books and learn the relationship between actions in the kitchen and the recipes. It might need descriptive word tr5ees to encode the kitchen actions, a sort of aid. But assume not, assume that it has mechanical sequences that match each recipe. It should be able to read the recipe book and identify the action sequence, by entropy matching.
Can the bot discover a circle? Give it the XY coordinates, representative sample. It will reduce precision until a rank great than one encoding tree. It should dicover the tree is perfectly symmetric, and construct a code that is linear in angle, a circle. Given a set of unknown data, the bot builds the encoding tree, then likely bounces that graph against a set of common graphs in its base, to find out how unique it is.
No comments:
Post a Comment