def run_latent_data(base_path, trn_data, tst_data, outpath, name, fold):
trndata = ANN_Executioner_Helper.get_latent_data(base_path, trn_data['name_index'], trn_data['stress'], use_input_sensitivity=True, normalize=True)
tstdata = ANN_Executioner_Helper.get_latent_data(base_path, tst_data['name_index'], tst_data['stress'], use_input_sensitivity=True, normalize=True)
ANN_Executioner_Helper.run_nn_train_until_convergence(trndata, tstdata, outpath, name, fold)
pass
def run(fold_object_path, number_of_fold, outpath, feature, duration,
delta_bool, delta2_bool, latent_data_path):
print 'Start at Fold object path : {}'.format(fold_object_path)
for i in range(number_of_fold):
trn_obj, tst_obj = ANN_Executioner_Helper.get_train_and_test_fold(
fold_object_path, number_of_fold, i)
trn_data = ANN_Executioner_Helper.get_data_from_obj(
trn_obj, feature, duration, get_only_stress=None)
tst_data = ANN_Executioner_Helper.get_data_from_obj(
tst_obj, feature, duration, get_only_stress=None)
trn_data_set = ANN_Executioner_Helper.get_ClassificationDataSet(
trn_data['Y'], trn_data['stress'])
tst_data_set = ANN_Executioner_Helper.get_ClassificationDataSet(
tst_data['Y'], tst_data['stress'])
print "Number of training patterns: ", len(trn_data_set)
print "Input and output dimensions: ", trn_data_set.indim, trn_data_set.outdim
print 'Plain data : '
ANN_Executioner_Helper.run_nn(trn_data_set, tst_data_set, outpath,
'Plain_data', i)
print 'Latent data : '
run_latent_data(latent_data_path, trn_data, tst_data, outpath,
'Latent_data', i)
print
def run_latent_data(base_path, trn_data, tst_data, outpath, name, fold):
trndata = ANN_Executioner_Helper.get_latent_data(
base_path,
trn_data['name_index'],
trn_data['tone'],
use_input_sensitivity=True,
normalize=False)
tstdata = ANN_Executioner_Helper.get_latent_data(
base_path,
tst_data['name_index'],
tst_data['tone'],
use_input_sensitivity=True,
normalize=False)
ANN_Executioner_Helper.run_nn(trndata, tstdata, outpath, name, fold)
pass