Principles of Data Science - Second Edition

Neural networks

Probably one of the most talked about machine learning models, neural networks are computational networks built to model animals' nervous systems. Before getting too deep into the structure, let's take a look at the big advantages of neural networks.

The key component of a neural network is that it is not only a complex structure, but it is also a complex and flexible structure. This means the following two things:

• Neural networks are able to estimate any function shape (this is called being non-parametric)
• Neural networks can adapt and literally change their own internal structure based on their environment

Basic structure

Neural networks are made up of interconnected nodes (perceptrons) that each take in input (quantitative value), ...