You're looking at Less Wrong's discussion board. This includes all posts, including those that haven't been promoted to the front page yet. For more information, see About Less Wrong.

Username comments on Open thread, Jan. 26 - Feb. 1, 2015 - Less Wrong Discussion

6 Post author: Gondolinian 26 January 2015 12:46AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Article Navigation

Comments (431)
Tags: open_thread

Comments (431)

You are viewing a single comment's thread. Show more comments above.

Comment author: Antisuji 26 January 2015 04:02:16AM 5 points [-]

Schmidhuber's formulation of curiosity and interestingness as a (possibly the) human learning algorithm. Now when someone says "that's interesting" I gain information about the situation, where previously I interpreted it purely as an expression of an emotion. I still see it primarily about emotion, but now understand the whys of the emotional response: it's what (part of) our learning algorithm feels like from the inside.

There are some interesting signaling implications as well.

Comment author: Username 26 January 2015 05:42:20AM *  4 points [-]

Direct link to the page on the theory.

That's really interesting! (ha!) I recommend reading the full page for good examples, but here's a summary:

Apart from external reward, how much fun can a subjective observer extract from some sequence of actions and observations? His intrinsic fun is the difference between how many resources (bits & time) he needs to encode the data before and after learning. A separate reinforcement learner maximizes expected fun by finding or creating data that is better compressible in some yet unknown but learnable way, such as jokes, songs, paintings, or scientific observations obeying novel, unpublished laws.

Powered by Reddit Powered by Reddit