Learning is interaction between learner and environment
Supervised
* Spam email detection
* Labelled data
Unsupervised
* No distinction between training and test data
* Clustering data into subsets of similar objects
Reinforcement learning
* Based on reward/punishment on the output result
Active learner
* Interacts with the environment at training time
* Pose queries or perform experiments
Passive learner
* Observes information provided by the environment
* Does not influence or direct it
Online vs Batch learning protocol
* The key distinction between online and batch learning is whether the learner must respond immediately during the learning process (online) or can first process large amounts of data before outputting conclusions (batch).