Compute mapping for x
Compute mapping for x
the input matrix. Before application, input should be summed already.
value of f(x)
Differentiate function at output value fx.
Differentiate function at output value fx.
Differentiation point vector, by output value.
Differentiation value of this function, generally squared matrix with same dimension.
Assumes CrossEntropyErr objective function, and in that case, CEE's derivative -1/out will be canceled with Softmax derivation term out(delta_ij-out)in. Hence, we designed softmax with derivation term (delta_ij-out)in.
Differentiate function at output value y.
Differentiate function at output value y.
Differentiation point vector, by output value.
Differentiation value of this function, generally squared matrix with same dimension.
Initialization range of weight
Initialization range of weight
the number of fan-in i.e. the number of neurons in previous layer
the number of fan-out i.e. the number of neurons in next layer
the initialized range of weight
Activation Function: Softmax function for CrossEntropyErr
We assumed the input of activation is a row vector.
We assumed the error function is CrossEntropyErr.