AIExplainer
Machine Learning Intermediate 2 min read

What is proxy labels?

Labels or annotations used as substitutes for true labels in machine learning models

Proxy labels are used when the true labels for a dataset are difficult or expensive to obtain. They can be used to train machine learning models, but may not be as accurate as using the true labels.

Think of proxy labels like using a map to navigate a new city instead of having a personal guide. The map can get you close to your destination, but it may not show you the exact best route.

For example, in a medical diagnosis application, proxy labels might be used to train a model to diagnose diseases based on symptoms, rather than using actual diagnoses from doctors.

Proxy labels are used in machine learning to train models when true labels are not available. They can be used in a variety of applications, including image classification, natural language processing, and recommender systems.

One common misconception is that proxy labels are just as good as true labels. However, proxy labels can be noisy or biased, which can affect the accuracy of the model.

The use of proxy labels has been around for several decades, but has become more popular in recent years with the rise of machine learning and deep learning.

surrogate labels pseudo-labels approximate labels

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