Rectified Linear Unit (ReLU)
Definition
ReLU sets negative inputs to zero and leaves positive inputs unchanged.
An activation function that sets negative inputs to zero and passes positive inputs unchanged.
Simply explained
A computational step that clips values below zero and passes the rest through.
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
Possible example: A network layer uses ReLU, setting negative inputs to zero and allowing positive values to pass unchanged.
Limits and pitfalls
- Ignoring dying ReLUs with permanently negative inputs or considering ReLU optimal for all architectures.
Origin & history
ReLU became widely established through work on deep neural networks around 2010.
Use in mid-sized companies
- Possible application: A network layer uses max(0,x) as a nonlinear activation.
Frequently asked questions
What does Rectified Linear Unit (ReLU) mean?
An activation function that sets negative inputs to zero and passes positive inputs unchanged. What mistakes should be avoided? Ignoring dying ReLUs with permanently negative inputs or considering ReLU optimal for all architectures.
Sources
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: