forward_backward_pass: Compute crucial quantities evaluated from one...

In billyhw/netzuko:

Compute crucial quantities evaluated from one forward-Backward pass through the neural network

Description

Compute crucial quantities evaluated from one forward-Backward pass through the neural network

Usage

forward_backward_pass(
  x,
  y,
  w,
  activation,
  output_type,
  dropout = FALSE,
  retain_rate = NULL,
  forward_only = FALSE
)

Arguments

x	The inputs
y	The outputs
w	The list of weights: 1st element are connection of weights from input to 1st hidden layer, and the last element are connection weights from the last hidden layer to the outputs
output_type	The output type: either "numeric" (regression) or "categorical" (prediction).
dropout	Boolean to indicate whether dropout is used
retain_rate	The proportion of units to retain during dropout
forward_only	If TRUE the function will only evaluate the forward pass

Value

A list containing the following elements: p: the output probabilities delta: a list of errors backpropagated throught the layers z: the hidden units values

billyhw/netzuko documentation built on March 23, 2022, 4:26 p.m.