Summary
This chapter mainly focused on artificial neural networks (the Multilayer Perceptron, in particular), which have become increasingly important in the field of machine learning due to their capability to tackle highly complex data problems that usually use extremely large datasets with patterns that are impossible to look at by the human eye.
The main objective is to emulate the architecture of the human brain by using mathematical functions to process data.
The process that is used to train an ANN consists of a forward propagation process, the calculation of a cost function, a back-propagation process, and the update of the different weights and biases that help to map the input values to an output.
In addition to the variables of the weights and biases, ANNs have multiple hyperparameters that can be tuned to improve the performance of the network, which can be done by modifying the architecture or training process of the algorithm. Some of the most popular hyperparameters...
