Scaled Exponential Linear Unit (SELU)
Definition
SELU helps neural networks maintain stable value ranges between their layers.
An activation function whose parameters promote self-normalizing networks under certain architectural and initialization conditions.
Simply explained
Like a built-in balance, SELU keeps the propagated numbers within a favorable range under suitable conditions.
Why it matters in practice
The term helps to select, develop, evaluate AI systems in a well-founded manner and realistically assess their technical risks.
Example from business
A fully connected network uses SELU with LeCun normal initialization and AlphaDropout.
Limits and pitfalls
- Do not promise self-normalization regardless of conditions or combine normal dropout unchanged.
Origin & history
Klambauer and co-authors published SELU in 2017.
Use in mid-sized companies
- Possible application: A fully connected network uses SELU with LeCun normal initialization and AlphaDropout.
Frequently asked questions
What does Scaled Exponential Linear Unit (SELU) mean?
An activation function whose parameters promote self-normalizing networks under certain architectural and initialization conditions. What mistakes should be avoided? Promising self-normalization regardless of conditions or combining normal dropout unchanged.
Sources
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: