Regularization

2020-01-28 本文已影响0人 spraysss

overfitting

如果特征过多，但是训练集不够时，很有可能会出现overfitting

解决overfitting的几种方法

减少特征数

手动选择需要保留哪些特征
Model selection algorithm

Regularization

cosf function
$J(\theta_0,\theta_1,...,\theta_n)=\frac{1}{2m}\sum_{i=1}^{m}(h_\theta(x^i)-y^i)^2+\frac{1}{2m}\lambda\sum_{j=1}^n\theta_j^2$

Keep all the features,but reduce magnitude/values of parameters $\theta_j$
it works well when we have a lot of features,each of which contributes a bit to predicting y

上一篇下一篇

猜你喜欢

热点阅读