Types of activation functions in Deep Learning

There are various aspects of deep learning that we usually have to consider while making a deep learning model. Choosing the right number of layers, the activation function, number of epochs, loss function, the optimizer to name a few. I am revisiting these concepts for one of my projects so I decided to write about the different activation functions we use. So why do we even use the activation function and not just feed the summation directly to the next layer. The problem if we do this would be that the layers of the neural network wont be able to learn complex functions over time. The activation function adds non linearity to the model.