The most widely used artificial intelligence algorithm today is the one called CNN (Convolutional Neural Networks). So what is this algorithm and how does it work? In the CNN algorithm, all layers work exactly connected to each other. The layers in the area have a filter feature and provide optimization in size and training time.

The most critical point of deep learning is AlexNet. In the above algorithm, random data is removed from the structure and the problem of memorizing the data of the model is avoided. The CNN algorithm can perform the tasks listed in the photo below flawlessly if you have enough data available. In the structure of CNN, the layers go even deeper. With much more filtering method, the desired result is obtained more clearly.

So what are the problems in the CNN algorithm? When we look at the photo below, our brain will realize that this photo is the statue of liberty, despite the fact that it may have been taken from different angles and in different shades of light. But the CNN algorithm does not have such a capability. When you promote one of these photos using the CNN algorithm, it probably won't recognize the other photos as the statue of Liberty. The Era of Capsule Networks Begins! Capsule networks are a network structure that is closer to the work of the human brain. The reason why it is called a capsule is because they are not connected to each other as in the CNN structure, but in a nested structure, they are named this way. He introduced the method of adding more than one capsule layer in one layer.

As a result, Capsule Networks are a deep learning (deep learning) structure that offers the opportunity to model the hierarchical structure in the image more efficiently. In the future, we are getting closer step by step to making machines that can learn like the human brain by improving the algorithms even more.