Deep Learning and its Applications using Python
by Niha Kamal Basha, Surbhi Bhatia Khan, Abhishek Kumar, Arwa Mashat
4Multilayer Perceptron
As discussed in first chapter on presentation of intelligent learning algorithms, multilayer perceptron is one of the supervised learning algorithm, which is an example of artificial neural network with feedback loop. Neural network [1] is a combination of different algorithms arranged in sequence to solve the problems by training and testing the networks with input. The process involved in this network resembles the working of human brain. The structure of neuron in the human brain is shown in Figure 4.1. The neurons cell body is soma. The chemical messages are received by dendrites which are extended from soma. Axon acts as a transmitter to transfer electro-chemical signal from one neuron to another and the myelin sheath acts as an insulator. Axon terminal (bouton) acts as a convertor to pass information from one neuron to another by converting electric signal into chemical signal. The same way the neural networks pass input and process them among multiple layers. Before getting into multilayer perceptron, the working of perceptron and artificial neural network has been given shown below.
4.1 Artificial Neural Network
It is a system of nodes in different layers [2] which interconnected to each other to process information. In Artificial Neural Network (ANN), the structure of the network is layered as input, hidden and output layer. In this, first input layer receives input in the form of text, image, number, audio files etc. The middle hidden layer ...