Deep learning and deep neural networks
Now, just shift from ANN to DNN. In the upcoming section, we will see deep learning, architecture of DNN, and compare the approaches of DL for NLP and ML for NLP.
Revisiting DL
We have seen some basic details about DL. Here, the purpose is just to recall things as a little refresher. ANN that is not two or three layers but many layers deep is called DNN. When we use many layers deep neural networks on lots of data using lots of computing power, we call this process deep learning.
Let's see the architecture of a deep neural network.
The basic architecture of DNN
In this section, we will see the architecture of a DNN. The pictorial representation looks very simple and is defined with some cool mathematical formulas in the form of activation function, activation function for hidden layer, loss function, and so on. In Figure 9.41, you can see the basic architecture of a DNN:
Figure 9.41: Architecture of DNN (Image credit: https://cdn-images-1.medium.com/max/800...
