Datalab Logo

Neural Networks

Neural Networks

Neural Networks


The human brain unconsciously processes millions of signals a day, making it one of the most advanced wonders of the world. Humans have an extraordinary learning ability, being able to connect pieces of information and to interpret visual and audio signals. Researchers used the brain as a source of inspiration during the development of artificial intelligence and came up with artificial neural networks.

A neural network is created with multiple artificial neurons (perform a calculation). A neural network can be used to perform a classification task or regression task. A simple example of classification is: predicting if someone is eligible for a bank loan. A regression task can be used to predict the price of a house (output layer), given some house features (input layer).

The goal of neural networks is to come up with an architecture which is able to learn the connection between the input variables (house features) and the output variable (house price). This connection is iteratively learned by evaluating the input and output for various of known cases.

Source:
McKinsey: “An executive’s guide to AI”


These known cases can be house properties where both the input and output is known and usually thousands of samples are required to learn the pattern. After the learning process is completed, the network is used on houses where only the features (input) is known but no house price is listed.