一、underfitting 和 overfitting 中，error的變化情況
IMG_20170425_135852.jpg
Underfitting:
training error will high, and cross validation error also will be high.
Overfitting:
training error will be low,but cross validation will be low.
二翻默、在regularization中，lambda對(duì)error的影響
IMG_20170425_144508.jpg
隨著lambda 的增加，training error在增加按声，整個(gè)過程由overfitting->fitting->underfitting，所以cross validation由大->小->大恬吕。
實(shí)驗(yàn)中签则，需要不斷調(diào)整lambda的值(一般從0逐漸增加,step可以為0.01，或者0.1等)铐料，最終找到error相對(duì)較小的lambda.
