Displaying 1 result from an estimated 1 matches for "node_eval".
Did you mean:
_node_val
2007 Jul 15
0
neural networks function in R
...on? Specifically, my question is that does my gradient function has to be flat representation of each parameter like the following arbitrary example:
2*w_i1_h1+3*w_h1_o1+4 (where w_i1_h1 is weight of input node 1 to hidden layer node 1, and vice versa)
Or I can do a nest expression like:
sigmoid(node_eval(input_x, j))
here my node_eval is a function that takes in input_x feature value vector and produces output to node j in the next layer.
If expression base evaluation like in case 2 is possible, it will significantly simplify the definition of my gradient function output.
Any help would be re...