Linear Basis Function Models

polynomial functions: changes in one region affect all other regions.
- spline functions : don't have to worry about globalness.
Gaussian functions : \(\exp\{-\frac{(x-\mu)^2}{2s^2}\}\)
sigmoidal functions : \(\sigma(\frac{x-\mu}{s})\), while \(\sigma(a)\) is the logistic sigmoid function defined by \(\sigma(a) = \frac{1}{1 + \exp(-a)}\).
- tanh functions : \(\tanh(a) = 2\sigma(a) - 1\).
Fourier basis : an expansion in sinusoidal functions.
- wavelets

Reference