Convolutional Neural Networks

Convolutional Neural Networks (CNNs) are used for amount of features that implements same transformation to different subsets of the features. CNNs are often used for images hence features are pixel values and a single convolution is a single subset of the image being transformed.

CNN implements the ideas of,

• Local connectivity: A hidden unit connecting to small patch of units in the previous layer

• Shared weights: Groups of hidden units share same set of input weights called mask, filters, or kernels.