Convolutional Neural Network
A deep learning model that automatically detects features like edges and patterns in images using convolutional layers, enabling it to recognize complex structures without manual input.
A convolutional neural network is a class of deep learning algorithms designed to process data with a grid-like topology, such as images. It uses convolutional layers to automatically detect features like edges and patterns, enabling it to recognize complex structures without manual feature engineering. The name derives from the mathematical operation of convolution applied to input data.
What this means in real life
When you upload a photo to your phone and it automatically tags your friend's face, a CNN is likely analyzing pixel patterns to recognize facial features—starting with simple edges, then combining them into eyes and noses, then whole faces.
What it isn’t
It is not a general-purpose neural network that works equally well on all data types. CNNs are specifically optimized for spatial, grid-structured data like images; they are inefficient and unnecessary for purely sequential text or tabular data without spatial relationships.
Commonly misused online
People often use 'CNN' to mean any deep learning model that 'sees' or processes images, when technically only architectures with convolutional layers qualify. Transformer-based vision models, for instance, process images without convolutions.