A neural network (as you may have guessed) is a machine learning model that is able to mimic the complex functionality of the human brain. Models such as these consist of nodes (or neurons) that are interconnected to process data, learn new patterns, and allow for decision making.

A neural network is able to identify patterns from a dataset without having pre-defined rules. They're built from five key components:

Neurons: Units that recieve inputs (has activation thresholds).

Connections: The links between all the neurons, carrying information and regulated by weights and biases.

Weights & Biases: Determines the strength of connections.

Propagation Functions: Helps to transfer data across layers of neurons.

Learning Rule: Adjusts weights and biases over time to decrease error.

Neural networks learn though being fed data. Based on the current parameters of the network an output is generated. Through Iterative Refinement the neural network will iteratively adjust its weights and balances, gradually improving its own performance.