# Convolutional Neural Network **Domain:** Machine Learning / Computer Vision **Doc Type:** Canonical Architecture Node **Maturity:** Foundational ## Definition A **convolutional neural network**, or **CNN**, learns spatially organized features through shared local transformations, layered representation and aggregation across neighboring positions. ## Historical Lineage [[wiki/Kunihiko Fukushima|Kunihiko Fukushima's]] [[wiki/Neocognitron|neocognitron]] established a direct architectural ancestor through alternating feature-sensitive and pooling cells that produced position-tolerant recognition. ## Technical Context Convolutional structure reuses detectors across a field, reducing parameter count while preserving local relationships. Pooling or related aggregation operations provide controlled tolerance to translation. ## See Also [[wiki/Computer Vision|Computer Vision]], [[wiki/Deep Learning|Deep Learning]], [[wiki/Representation Learning|Representation Learning]]