Lazy learning

Definition

Lazy learning refers to a form of machine learning where generalization beyond the training set is delayed until a query is made to the system. This is opposed to eager learning, where there is a preprocessing step where the training set is used to learn a functional form for the prediction function.

Situations where lazy learning is appropriate

• Lazy learning is suited for situations where the entire training set can be stored or can be quickly retrieved for each instance where a new prediction needs to be made. For situations where the training set is too large to store or retrieve for each instance of prediction, lazy learning is not possible.
• Lazy learning is best suited for situations where computing the value of the prediction function at one point requires only a strategically determined subset of the training set, rather than the whole training set, and/or situations where there are ways of computing the prediction function at a particular point that do not entail finding a closed form for the prediction function globally.