Where, is the number of outputs () and is the number of inputs
Key Model Selection Decisions:
- Number of RBF units (hidden neurons) - How many radial basis functions to use in the hidden layer
- RBF centers - Where to position each RBF unit in the input space (the centers of the Gaussian functions)
- RBF widths (σ or spread parameters) - The width/spread of each radial basis function, which controls how much of the input space each unit responds to
- Output weights () - The weights connecting the RBF units to the output layer