module

Ai4cr::NeuralNetwork::Cmn::MiniNetConcerns::TrainAndAdjust

Instance methods

calculate_error_distance
Source
derivative_propagation_function
Source
guesses_best
Source
init_net_re_train
Source
input_deltas
Source
input_deltas=(input_deltas)
Source
last_changes
Source
last_changes=(last_changes)
Source
load_outputs_deltas(outputs_deltas)
Source
load_outputs_expected(outputs_expected)
Source
output_deltas
Source
output_deltas=(output_deltas)
Source
output_errors
Source
output_errors=(output_errors)
Source
outputs_expected
Source
outputs_expected=(outputs_expected)
Source
set_deriv_scale_prelu(scale)

Per Learning Style:

Source
step_backpropagate
Source
step_calc_input_deltas

Calculate deltas for hidden layers

Source
step_calc_output_errors
Source
step_calculate_output_deltas

Calculate deltas for output layer

Source
step_load_chained_outputs_deltas(outputs_deltas)

This would be a chained MiniNet's input_deltas e.g.: mini_net_A feeds is chained into mini_net_B So you would mini_net_A.step_load_chained_outputs_deltas(mini_net_B.input_deltas)

Source
step_load_outputs(outputs_expected)
Source
step_update_weights(parallel = false)
Source
step_update_weights_v1
Source
step_update_weights_v2
Source
train(inputs_given, outputs_expected, until_min_avg_error = UNTIL_MIN_AVG_ERROR_DEFAULT)

training steps

TODO: utilize until_min_avg_error

Source