Max Pooling in Convolutional Neural Network(CNN)

Original article was published by Manik Soni on Deep Learning on Medium

Max Pooling in Convolutional Neural Network(CNN)

Max Pooling is one of the steps in building a Convolutional Neural Network(CNN)

Max Pooling helps to reduce the feature map in order to do the classification more precisely. Let’s take an example to understand this topic better.

Example: Consider the ‘cheetah’ image.

You can take different side video of different ‘cheetahs’

But we want the machine to understand in the same way, that is extracting a common feature from already extracted feature map which helps to do the categorization that the given animal is a ‘cheetahs’

This concept is known as Max Pooling.

How does Max Pooling work?

Let us suppose there is a Feature Map,

Now, we are doing the pooling of the above feature.

Now, Reducing the Feature Map into Pooled Feature Map. Make a 2*2 matrix and take a maximum from the 2*2 matrix.

Again, mapping the pooled feature

Do this mapping until the end.

Fill the Pooled feature map until the end.

Now each of the feature maps, we are applying max-pooling to get a pooled feature map.

So in the real-life application if, we try to visualize how pooling is applied at the back of Convolutional Neural Network(CNN)?.