Is there predictive knowledge for which there is no useful explanation, other than knowledge which codifies physically stochastic processes?
Conversation
Replying to
1
That’s interesting. Will ponder.
1
Show replies
Replying to
regular verbs, I mean. There are rules, they work, they predict.
Replying to
Evolutionary programming (genetic algorithms, etc.) often generates predictive functions that are incomprehensible.
1
Yeah, but there exist useful explanations for many of those results—the GAs just don’t surface them.
1
Show replies
Replying to
It sounds like you’re asking about Kolmogorov complexity. Low complexity => mechanically explainable, maybe generalizable?
1
2
1
2
Show replies