Empirical Risk Minimization

In Machine Learning, the training set error is often called the empirical error or empirical risk, and this is the error the classifier incurs over the sample:

$L_{S} (h) = \frac{| {i \in [m] : h (x_{i}) \neq y_{i}}}{m}$

Given a hypothesis class $H$ , finding the hypothesis $h \in H$ that minimizes the empirical risk is a simple learning strategy.

Jethro's Braindump

Empirical Risk Minimization

Links to this note