True Risk

machine-learning

Definition

True Risk

True risk $R(h)$, $h$ being a hypothesis or model is the expected loss on all possible data, including future, unseen examples. In practice, $R(h)$ is incomputable because the true distribution $\mathcal{D}$ is unknown.

$$R_{\mathcal{D}}(h)={\mathbb{E}}_{(x,y)\sim \mathcal{D}}[\mathcal{L}(h;(x,y))]$$

Note that this is for a supervised setting.