assembly.jc 2019-5-18 00:06
Probably approximately correct (PAC) learning，幾年前學過，舊事重提是因為有人問起 Understanding Machine Learning: From Theory to Algorithms, Ex 2.4, 第二題 (P. 41):
小弟的理解是 L(D,f)(h) 是 Bernoulli distribution，E [Ls (h)] 只是 binomial distribution 的 expected value，即係 mp / m = p = L(D,f)(h)。但書入面無提及 Bernoulli distribution，所以其他人唔係好同意，有無熟概率嘅 ching 可以講下？
[[i] 本帖最後由 assembly.jc 於 2023-10-22 03:34 編輯 [/i]]