Unsupervised Learning

When red teaming AI systems using unsupervised learning (e.g., clustering algorithms), focus areas include assessing whether the discovered patterns reveal sensitive information, if the model can be manipulated to group data incorrectly, or if dimensionality reduction techniques obscure security-relevant features. Understanding these models helps identify risks associated with pattern discovery on unlabeled data.

Learn more from the following resources:

• @article@How Unsupervised Learning Works with Examples
• @article@Supervised vs. Unsupervised Learning: Which Approach is Best?