Machine Learning

Machine Learning is the branch of Artificial Intelligence in which computers gain the ability to learn from data — using statistical techniques — without being explicitly programmed by a human being for every rule. Instead of a programmer writing down all the logic, the system works the patterns out from data on its own.

Models: algorithms that learn

Machine Learning is built from algorithms that use data to learn on their own and make predictions. These algorithms are called models, and they go through a standard workflow:

1. Training — the model is first trained on training data, from which it learns patterns.
2. Testing — the model is then tested on separate testing data to check how well it has learned.
3. Successive trainings — training is repeated until the model's results reach an acceptable level of accuracy.
4. Prediction — once accurate enough, the model is put to work making predictions about new and unknown data it has never seen before.

The key ideas in one place

• Machine Learning is a subsystem of AI — every ML system is an AI system, but AI also includes other approaches.
• The defining feature is learning from data rather than explicit programming.
• Training data teaches the model; testing data measures it; only then does it face real, unseen data. …