Empirical Risk MinimizationA generalization of the Maximum Likelihood principle MLE: replace the log likelihood with any other loss function lllL(θ)=1n∑i=1nl(Yi,θ)\mathcal{L}(\theta) = \frac{1}{n} \sum_{i=1}^n \mathcal{l}(Y_i, \theta)L(θ)=n1∑i=1nl(Yi,θ)ERM Notion