Give a class $\mathcal{H}$ of predictors from the set X to $\{0,1\}$. Here each $h \in \mathcal{H}$ is a function from $h: X \longrightarrow \{0,1\}$. The Empirical Risk Minimization is defined as

$$ERM_{\mathcal{H}} \in Arg min_{h \in \mathcal{H}} (L_s(h))$$ Where $S=\{(x_1,y_1,\cdots,(x_m,y_m))\}$ is a sample $L_S(h)$ denotes the loss of hypothesis $h$ on the sample $S$.

The goal of the learner is to find an $h$ with as small loss as possible.

In the above it is paradigm not an algorithm. I am not getting the point why it is not referred as algorithm?


Video Link


1 Answer 1


A paradigm is an abstract concept. It is a general method useful to know to solve a problem or construct algorithms, but not a particular algorithm. In your example, the paradigm is to minimize the empirical risk. But it does not say how to perform the minimization concretely.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.