Convolutional Neural Networks (CNN)
Convolutional neural networks (CNN)
are one of the variants of neural networks that are heavily used in the field of computer vision. It gets its name from the type of hidden layers. The hidden layers of a CNN generally consist of convolutional layers, pooling layers, fully connected layers, and normalization layers. Here it simply means that instead of using the usual activation functions mentioned above, convolution and pooling functions are used as activation functions.
Image credits: Taken from a Google search, source was not mentioned. let me know if it's your image and i can add credit
To understand this in detail one needs to understand what convolution and pooling are. Both of these concepts are borrowed from the field of computer vision and are defined below.
Conversation: Conversation operates on two signals (in 1D) or two images (in 2D): you can think of one as an "input" signal (or image), and the other (called a kernel). Input image as "filter", producing an output image (so convolution takes two images as input and produces a third one as output). [5]
In layman terms, it takes an input signal and applies a filter to it, essentially multiplying the input signal with the kernel to obtain the modified signal. Mathematically, a determination of two functions f and g is defined as
Which is nothing but the dot product of the input function and the kernel function.
In the case of image processing, it is easy to imagine a kernel as sliding over the entire image and thus changing the value of each pixel in the process.
Image credits: Taken from a Google search, source was not mentioned. let me know if it's your image and i can add credit
Image credits: machine learning gurus
Pooling: Pooling is a sample-based rationalization process. The objective is to sample an input representation (image, hidden-layer output matrix, etc.), reducing its dimensionality and allowing for assumptions about the features contained in the subfields.
There are 2 main types of pooling commonly known as maximum and minimum pooling. As the name suggests maximum pooling is based on taking the maximum value from the selected region and min pooling is based on taking the minimum value from the selected region.
Thus, as one can see a Convolutional Neural Network or CNN is basically a deep neural network that consists of hidden layers in the convolution and pooling functions in addition to the activation function to introduce non-linearity.