AI engineering glossary · Deep learning core
What is huber loss?
Huber loss (also called Smooth L1 loss): A regression loss that acts like squared error for small mistakes and like absolute error for large ones, switching at a threshold delta. It stays smooth near zero yet is less thrown off by outliers than plain squared error.
Huber loss: an example
With delta = 1, an error of 0.5 costs 0.125 while an error of 10 costs 9.5, not 50.
Related terms
- Loss function
- Gradient descent
- Exploding gradient
More Deep learning core terms
- Input layer
- Instance normalisation
- Iteration
- Label smoothing
- LAMB optimiser
- Layer
- Layer normalisation
- Leaky ReLU
Where you learn it
Huber loss is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.