AI engineering glossary · Deep learning core
What is label smoothing?
Label smoothing: A regularisation trick that replaces hard 0/1 target labels with slightly softened ones, such as 0.9 for the true class and the remaining 0.1 spread over other classes. It discourages overconfident predictions and can improve calibration.
Label smoothing: an example
With 10 classes and smoothing 0.1, the target for the right class is about 0.91 and each wrong class gets 0.01.
Related terms
More Deep learning core terms
- LAMB optimiser
- Layer
- Layer normalisation
- Leaky ReLU
- Learning rate finder
- LeCun initialisation
- Linear scaling rule
- Lion optimiser
Where you learn it
Label smoothing is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.