#empirical-risk

Question 1: Single choice

mediumAI-checkedML Fundamentals

How does the empirical risk of a model differ from its population risk?

Empirical risk averages the loss over the finite training set; population risk is the loss expected under the true data distribution1
Empirical risk is the loss expected under the true data distribution; population risk averages the loss over the finite training set2
Empirical risk is the mean loss on the test set; population risk is the mean loss on the training set3
Empirical risk adds a regularization penalty to the loss; population risk is the plain data loss on the same examples4