def _two_label_performance(target_names, params):
# get_params
noise_amount = params['noise_amount']
# set params
params['target_names'] = target_names
print_params(**params)
# fit hopfield
print('\n.. fitting hopfield\n')
hf, X, y, target_names, params = fit_hopfield(params)
print_params(**params)
# recall
print('\n.. recalling\n')
X, X_noise, X_recall = recall_with_noise(clf=hf, X=X,
noise_amount=noise_amount)
print_header('result')
similarities, accurate = get_recalling_performance(X, X_recall)
print('similarity:', np.mean(similarities))
print('accuracy:', np.mean(accurate))
similarity = np.mean(similarities)
accuracy = np.mean(accurate)
return similarity, accuracy
def hopfield_single_performance(
n_sample,
n_label,
noise_amount,
fit_mode,
save_fig,
):
# parameters
params = {
'n_sample': n_sample,
'n_label': n_label,
'noise_amount': noise_amount,
'fit_mode': fit_mode,
}
print_params(**params)
# fit hopfield
print('\n.. fitting hopfield\n')
hf, X, y, target_names, params = fit_hopfield(params)
print_params(**params)
# recall
print('\n.. recalling\n')
X, X_noise, X_recall = recall_with_noise(clf=hf, X=X,
noise_amount=noise_amount)
print_header('result')
similarities, accurate = get_recalling_performance(X, X_recall)
print('similarity:', np.mean(similarities))
print('accuracy:', np.mean(accurate))
# compare 3 images & save
if save_fig:
print('\n.. view recalling result\n')
view_recalling_result(X, X_noise, X_recall,
accurate=accurate, **params)
similarity = np.mean(similarities)
accuracy = np.mean(accurate)
return similarity, accuracy