#print(model.clf)
available_perturbations = {
'rotations': rotation_perturbations(probs, num_repetitions),
'noise': noise_perturbations(probs, num_repetitions)
}
random.shuffle(available_perturbations['rotations'])
random.shuffle(available_perturbations['noise'])
train_perturbations = []
test_perturbations = []
for a, b in zip(available_perturbations['rotations'],
available_perturbations['noise']):
train_perturbations.append(Mixture([a, b]))
random.shuffle(available_perturbations['rotations'])
random.shuffle(available_perturbations['noise'])
for a, b in zip(available_perturbations['rotations'],
available_perturbations['noise']):
test_perturbations.append(Mixture([a, b]))
performance_predictor = train_performance_predictor(
test_data, y_test, threshold, train_perturbations, model, scoring)
evaluate_predictor(target_data, y_target, threshold,
test_perturbations, model, performance_predictor,
scoring, scoring_name, dataset_name,
'largeconvnet_mixture', learner_name,