Machine learning is a part of artificial intelligence in which computers make decisions based on data. Instead of relying only on explicitly specified rules, a computer uses patterns in data to make predictions or solve problems. section 1.2
From experience to prediction
To guess whether it will rain, recall past weather, form a rule such as “it rains here most of the time,” then use that rule to predict today’s weather. Machine learning follows a similar remember-formulate-predict process: it uses past data to build a model, then uses the model to make predictions. section 1.3
The type of machine learning depends partly on the data and the task:
- Supervised learning uses labeled data to predict labels; regression predicts a number, while classification predicts a type or class. section 2.1 section 2.5
- Unsupervised learning uses unlabeled data to simplify or group it. Reinforcement learning involves an agent navigating an environment toward a goal. section 2.5
Applications include predicting house prices, identifying spam, recognizing images, recommending movies, and diagnosing patients. The same broad approach can address problems in many fields. section 1.0 section 2.0
