Novel Class Discovery
Definition
Novel class discovery searches for unknown groups in unlabeled data.
A learning problem where new, previously unannotated classes are to be detected and grouped in unlabeled data.
Simply explained
A system finds, for example, a new error type for which there was previously no specific label.
Why it matters in practice
Helps realistically assess model quality, limits, and technical decisions.
Example from business
A quality model discovers a new error type in previously unlabeled images.
Limits and pitfalls
- Automatically labeling clusters as genuine classes or mixing known and new classes.
Origin & history
Novel Class Discovery evolved from clustering, semi-supervised learning, and open-world recognition.
Use in mid-sized companies
- Possible example: A quality model discovers a new error type in previously unlabeled images.
Frequently asked questions
What does novel class discovery mean?
A learning problem where new, previously unannotated classes are to be detected and grouped in unlabeled data.
Which mistakes should be avoided?
Automatically labeling clusters as genuine classes or mixing known and new classes.
Sources
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: