COMP 102 Lecture Notes - Lecture 18: Overfitting, Weighted Arithmetic Mean, Microsoft Onenote
Document Summary
Comp 102 lecture 18: machine learning cont. Midterm review: selecion sort- irst index; and not sliding down but swapping, runime. Machine learning make a computer learn with from data only: a data set, decision tree. A node a pariion of the input space. Internal nodes tests on each atribute don"t need to be binary. Leaf nodes also include the set of training examples that saisfy the test along the branch: algorithm of learning decision trees. A good test should provide informaion about the class label: informaion theory. Informaion content: let e be an event that occurs with probability p(e). Entropy: we have an informaion source s which emits symbols from an alphabet {s1 , . sk } with probabiliies {p1 , . pk }, each emission is independent of others. The average amount of info we expect from s outcome: call this entropy of s, note that this depends only on the probability distribuion, and not on the actual alphabet.