Max Pooling, Why use it and its advantages.

There are many operations that are applied to an image in order to extract the most important features using Convolution and Max Pooling.

Max Pooling is an operation that is used to downscale the image if it is not used and replace it with Convolution to extract the most important features using, it will take high computational cost. So we use Max Pooling to extract maximum value from the Feature map according to filter size and strides.

Max Pooling operation is always done after Convolution (Credit: Codicals)

We must use Max Pooling in those cases where the size of the image is very large to downsize it. Max pooling stores only pixels of the maximum value. These values in the Feature map are showing How important a feature is and its location. So, taking only the maximum value means extracting the most important feature in a region.

Two reasons for applying Max Pooling :

1. Downscaling Image by extracting most important feature

2. Removing Invariances like shift, rotational and scale

We must be thinking that Is downscaling the images is the only use of it. I would say no. There are many advantages of using Max Pooling over other Pooling operations (Min Pooling and Average Pooling).

Max Pooling is advantageous because it adds translation invariance. There are following types of it

1. Shift Invariance(Invariance in Position)

2. Rotational Invariance(Invariance in Rotation)

3. Scale Invariance(Invariance in Scale(small or big))

Different Invariances added by Max Pooling (Credit: Codicals)

You must be thinking What is the importance of adding invariance in images.

Here are its reasons.

1. Invariance in images is important if we care about whether a feature is present rather than exactly where it is.
2. In other contexts, it is more important to preserve the location of a feature.

Here is an example to visualize invariance. (Credit: Codicals)

I hope all your concepts are cleared. I clearly understood this topic Max Pooling its uses and advantages.

Stay tuned and Keep Learning.