the only useful algorithm in the brain is nearest neighbor search
1. you can almost always con anyone into believing that you understand stuff by employing some variant of nearest neighbor search and yapping away nearest pattern matches to the topic of interest
2. it is of tremendous use in quizzes, trivia contests, examinations etc -- for most conceivable metrics used for measuring your "intelligence" a nearest neighbor algorithm suffices either for getting the answer or more importantly for convincing the questioner of your competence to answer to the question
3. it is amazingly useful for supposedly high intelligence examinations involving complex formulas
the only hard part then is building up a knowledge base and an appropriate set of distance metrics on which to operate
Subscribe to:
Post Comments (Atom)
3 comments:
Don't you think amassing the knowledge in the neighborhood around every possible topic is an onerous task for a lifetime? Or do you believe the graph that encompasses all knowledge is so dense that the cardinality of the subset of vertices V, such that V union N(V) = universe, is extremely small?
Atleast I don't think the latter is true.
well .. the hypothesis is that for all practical purposes u need to only build a small knowledge base on which to do the lookup (say a small sphere of influence/knowledge)
agree that the entire knowledge base is pretty vast ..
my argument is that rather than try to give an absolutely perfect answer a nearest neighbor usually suffices
That's fine. The point I was trying to make is that even if it suffices to find the nearest neighbor, unless you know a lot, there will always be some topics such that the nearest neighbor to that topic in your domain of knowledge will be pretty unrelated to the actual topic!
In other words, for every topic, if the nearest neighbor in your domain of knowledge to that topic is actually related to the topic in some way, then you do know a lot.
Post a Comment