params_DPAv4_HW.txt

# NN params
input_size	50  # Generated
output_size	9  # Generated
epochs_num	30
train_patience    5
convergence_delta   1e-2
batch_size	256
normalize_features	true
shuffle_batches	false
batch_norm true
learning_rate	5.0E-4
decay_rate	0.99
decay_step	1
regularization_coef	1.0E-4
dropout_keep_prob	1.0
seed	42
name	noiseless_all_training_9class  # Generated
train_percentage	0.8
architecture	fc(50)-fc(50)-fc(50)

# GA params
algorithm GENERATION_TABOO
algo_seed	{42,43,44,45,46,47,48,49,50,51}
population_size	12
mutation_prob	0.3
elitism	true
taboo_size	12
taboo_attempts	1000
worker_num	1
node_set	ALL

# Stop conditions:
max_iterations	30
min_fitness	-1.0

# GA operators
crx.swap_subtree	    1
crx.swap_nodes		    1
crx.swap_constants	    1
crx.mean_constants	    1
crx.return_random	    1
mut.insert_root		    1
mut.insert_terminal	    1
mut.random_constant_set	1
mut.random_constant_set.min	-1.0
mut.random_constant_set.delta	2.0
mut.replace_node	    1
mut.remove_unary	    1
mut.remove_root		    1
mut.swap_order		    1

# Dataset
train_path	res/DPAv4/DPAv4_traces_train_HW.csv ; res/DPAv4/DPAv4_labels_train_HW.csv
test_path	res/DPAv4/DPAv4_traces_test_HW.csv ; res/DPAv4/DPAv4_labels_test_HW.csv
experiment_name	test