# Rectified Linear Unit **Entity class:** Concept or analytic term **Domain:** Machine Learning / Neural Networks **Doc Type:** Technical Concept Node **Maturity:** Foundational ## Definition A **rectified linear unit**, or **ReLU**, is an activation function that outputs zero for negative inputs and passes positive inputs linearly. ## Historical Context [[wiki/Jürgen Schmidhuber|Jürgen Schmidhuber]] credits a 1969 multilayer network by [[wiki/Kunihiko Fukushima|Kunihiko Fukushima]] with the earliest use of an activation now described as rectified linear. That historical attribution should remain identified as Schmidhuber's reading. ## See Also [[wiki/Artificial Neural Networks|Artificial Neural Networks]], [[wiki/Deep Learning|Deep Learning]], [[wiki/Gradient Descent|Gradient Descent]] ## Relationships <!-- BEGIN HUMANIZED RELATIONSHIPS 2026-09-11 --> This entry's documented connections are expressed in its definition and related-work routes, with provenance retained in the source-linked material. <!-- END HUMANIZED RELATIONSHIPS 2026-09-11 -->