Examples of form of f()

The space of functions f() that a learning algorithm explores defines the type of algorithm

• Nearest Neighbour: localized decision regions from lookup table.
• Perceptron (or linear classifier): linearly-separable decision regions
• Neural Networks: non-linear(but continuous) model of decision regions
• Classification rules: same as trees but need not be representable as a single tree (larger space)
• Bayesian Network: Representation of joint probability density on f()