COMP 4106 Study Guide - Comprehensive Final Exam Guide - Algorithm, Time, Heuristic
Document Summary
If judge cannot reliably tell the machine from the human, the machine. If want to distinguish between cancer (orange) and non-cancer (blue) If white blood cell count is high, then likely have cancer: also, could maybe tell from size of lymph nodes, measurements like above and are called feature x2. Training sample that we know have cancer g(x) < 0. Testing sample that we do(cid:374)"t k(cid:374)o(cid:449) if ha(cid:448)e (cid:272)a(cid:374)(cid:272)er. White blood cell count x1: classifier, 1. Compute g(x: how good is g(x)? (*testing data, testing data consists people who you know their identity (have cancer or not). Use to tell how good your classifier is: find optimal classifier, classifier is not definitive. Does not need to say yes or no, can say does not know. Bad if wrongly say no you do not have cancer but do: neural networks, brai(cid:374)"s (cid:373)odel is (cid:272)o(cid:373)pletel(cid:455) differe(cid:374)t tha(cid:374) (cid:272)o(cid:373)puter.