all_gBest_metrics = np.zeros((number_runs, 2))
runs_time = []
all_gbest_par = []
best_gBest_acc = 0
for i in range(number_runs):
print("Run number: " + str(i))
start_time = time.time()
pso = psoCNN(dataset=dataset,
n_iter=number_iterations,
pop_size=population_size,
batch_size=batch_size_pso,
epochs=epochs_pso,
min_layer=min_layer,
max_layer=max_layer,
conv_prob=probability_convolution,
pool_prob=probability_pooling,
fc_prob=probability_fully_connected,
max_conv_kernel=max_conv_kernel_size,
max_out_ch=max_conv_output_channels,
max_fc_neurons=max_fully_connected_neurons,
dropout_rate=dropout)
pso.fit(Cg=Cg, dropout_rate=dropout)
print(pso.gBest_acc)
# Plot current gBest
matplotlib.use('Agg')
plt.plot(pso.gBest_acc)
plt.xlabel("Iteration")
def pso_train(self, ui):
# Custom changes to psoCNN:
# within this function: cfg.epochs_pso set to 10 as per original paper for optimal performance
# within particle.py module, an argument verbose=2 was added to .fit call in model_fit and model_fit_complete
# function which uses pso to train on either the fashion_mnist or cifar10 datasets
# one hot encoding, using to_categorical is carried out within pso library (pso_cnn.py)
if ui.dataset == "fashion-mnist":
dataset = "fashion-mnist"
elif ui.dataset == "cifar10":
dataset = "cifar10"
# Algorithm parameters #
number_runs = ui.pso_runs # 10 runs is often used as default
number_iterations = ui.pso_iter # 10 runs is often used as default
population_size = ui.pso_pop # 20 runs is often used as default
# Run the algorithm #
results_path = "./results/" + dataset + "/"
if not os.path.exists(results_path):
os.makedirs(results_path)
all_gBest_metrics = np.zeros((number_runs, 2))
runs_time = []
all_gbest_par = []
best_gBest_acc = 0
for i in range(number_runs):
print("Run number: " + str(i))
start_time = time.time()
new_run = datetime.now()
print(new_run)
pso = psoCNN(dataset=dataset, n_iter=number_iterations, pop_size=population_size, \
batch_size=cfg.batch_size_pso, epochs=cfg.epochs_pso, min_layer=cfg.min_layer,
max_layer=cfg.max_layer, \
conv_prob=cfg.probability_convolution, pool_prob=cfg.probability_pooling, \
fc_prob=cfg.probability_fully_connected, max_conv_kernel=cfg.max_conv_kernel_size, \
max_out_ch=cfg.max_fully_connected_neurons, max_fc_neurons=cfg.max_fully_connected_neurons,
dropout_rate=cfg.dropout)
pso.fit(Cg=cfg.Cg, dropout_rate=cfg.dropout)
print(pso.gBest_acc)
# Plot current gBest
matplotlib.use('Agg')
plt.plot(pso.gBest_acc)
plt.xlabel("Iteration")
plt.ylabel("gBest acc")
plt.savefig(results_path + "gBest-iter-" + str(i) + ".png")
plt.close()
print('gBest architecture: ')
print(pso.gBest)
np.save(
results_path + "gBest_inter_" + str(i) + "_acc_history.npy",
pso.gBest_acc)
np.save(
results_path + "gBest_iter_" + str(i) +
"_test_acc_history.npy", pso.gBest_test_acc)
end_time = time.time()
running_time = end_time - start_time
runs_time.append(running_time)
# Fully train the gBest model found
n_parameters = pso.fit_gBest(
batch_size=cfg.batch_size_full_training,
epochs=cfg.epochs_full_training,
dropout_rate=dropout)
all_gbest_par.append(n_parameters)
# Evaluate the fully trained gBest model
gBest_metrics = pso.evaluate_gBest(
batch_size=cfg.batch_size_full_training)
if gBest_metrics[1] >= best_gBest_acc:
best_gBest_acc = gBest_metrics[1]
# Save best gBest model
best_gBest_yaml = pso.gBest.model.to_yaml()
with open(results_path + "best-gBest-model.yaml",
"w") as yaml_file:
yaml_file.write(best_gBest_yaml)
# Save best gBest model weights to HDF5 file
pso.gBest.model.save_weights(results_path +
"best-gBest-weights.h5")
all_gBest_metrics[i, 0] = gBest_metrics[0]
all_gBest_metrics[i, 1] = gBest_metrics[1]
print("This run took: " + str(running_time) + " seconds.")
# Compute mean accuracy of all runs
all_gBest_mean_metrics = np.mean(all_gBest_metrics, axis=0)
np.save(results_path + "/time_to_run.npy", runs_time)
# Save all gBest metrics
np.save(results_path + "/all_gBest_metrics.npy", all_gBest_metrics)
# Save results in a text file
output_str = "All gBest number of parameters: " + str(
all_gbest_par) + "\n"
output_str = output_str + "All gBest test accuracies: " + str(
all_gBest_metrics[:, 1]) + "\n"
output_str = output_str + "All running times: " + str(
runs_time) + "\n"
output_str = output_str + "Mean loss of all runs: " + str(
all_gBest_mean_metrics[0]) + "\n"
output_str = output_str + "Mean accuracy of all runs: " + str(
all_gBest_mean_metrics[1]) + "\n"
print(output_str)
with open(results_path + "/final_results.txt", "w") as f:
try:
print(output_str, file=f)
except SyntaxError:
print >> f, output_str
return