Machine Learning & Training

What Is Contrastive Learning?

Contrastive learning is a technique that trains models to produce representations where similar examples are close together and dissimilar ones are far apart. It is often used in self-supervised learning to learn useful features without labels. The approach has been influential in both vision and language representation learning.