initzero :: Functions (Neural Network Toolbox™)

initzero

Purpose

Zero weight and bias initialization function

W = initzero(S,PR)
b = initzero(S,[1 1])

initzero(S,PR) takes two arguments,

S
Number of rows (neurons)

PR
R x 2 matrix of input value ranges = [Pmin Pmax]

`S`	Number of rows (neurons)
`PR`	`R` x `2` matrix of input value ranges = `[Pmin Pmax]`

and returns an S x R weight matrix of zeros.

initzero(S,[1 1]) returns an S x 1 bias vector of zeros.

Here initial weights and biases are calculated for a layer with two inputs ranging over [0 1] and [-2 2] and four neurons.

W = initzero(5,[0 1; -2 2])
b = initzero(5,[1 1])

You can create a standard network that uses initzero to initialize its weights by calling newp or newlin.

To prepare the weights and the bias of layer i of a custom network to be initialized with midpoint,

Set net.initFcn to 'initlay'. (net.initParam automatically becomes initlay's default parameters.)
Set net.layers{i}.initFcn to 'initwb'.
Set each net.inputWeights{i,j}.initFcn to 'initzero'. Set each net.layerWeights{i,j}.initFcn to 'initzero'. Set each net.biases{i}.initFcn to 'initzero'.

To initialize the network, call init.

See newp or newlin for initialization examples.