AIExplainer
Machine Learning Intermediate 2 min read

What is empirical risk minimization?

A method used in machine learning to find the best model by minimizing the average error on a given dataset

Empirical risk minimization is a way to train machine learning models by trying to minimize the difference between the model's predictions and the actual outcomes in the training data. This is done by adjusting the model's parameters to reduce the average error

Empirical risk minimization is like trying to find the best route to a destination by minimizing the total distance traveled. Just as you would take the shortest path, a machine learning model tries to find the best parameters to minimize the total error

For example, a company might use empirical risk minimization to train a model to predict customer churn. The model would be trained on a dataset of customer information and the goal would be to minimize the error between the predicted churn and the actual churn

Empirical risk minimization is used in supervised learning to train models such as linear regression, logistic regression, and neural networks. It is also used in other areas of machine learning, such as deep learning and natural language processing

One common misconception is that empirical risk minimization is the same as minimizing the true risk, which is the error on the entire population. However, empirical risk minimization only minimizes the error on the training data, which may not be representative of the entire population

The concept of empirical risk minimization was first introduced by Vladimir Vapnik in the 1960s and has since become a fundamental concept in machine learning

empirical risk risk minimization error minimization

Three products for different needs — explore what’s relevant to you.