def train_cvrnn_model(model, sess, config, input_data, target_data,
trace_lengths, selection_matrix):
[
output_x,
# total_loss,
kl_loss,
likelihood_loss,
_
] = sess.run(
[
model.output,
# model.cost,
model.kl_loss,
model.likelihood_loss,
model.train_general_op
],
feed_dict={
model.input_data_ph: input_data,
model.target_data_ph: target_data,
model.trace_length_ph: trace_lengths,
model.selection_matrix_ph: selection_matrix
})
output_decoder = []
kld_all = []
ll_all = []
for batch_index in range(0, len(output_x)):
output_decoder_batch = []
for trace_length_index in range(0, config.Learn.max_seq_length):
if selection_matrix[batch_index][trace_length_index]:
output_decoder_batch.append(
output_x[batch_index][trace_length_index])
kld_all.append(kl_loss[batch_index][trace_length_index])
ll_all.append(likelihood_loss[batch_index][trace_length_index])
else:
output_decoder_batch.append(
np.asarray([0] * config.Arch.CVRNN.x_dim))
output_decoder.append(output_decoder_batch)
output_decoder = np.asarray(output_decoder)
acc = compute_rnn_acc(output_prob=output_decoder,
target_label=target_data,
selection_matrix=selection_matrix,
config=config)
print_number = random.randint(1, 21)
if print_number == 1:
print("acc is {0} with mean KLD:{1} and mean ll:{2} ".format(
acc,
np.mean(kl_loss),
np.mean(likelihood_loss),
))
# print acc
# if cost_out > 0.0001: # TODO: we still need to consider how to define convergence
# converge_flag = False
cost_out = likelihood_loss + kl_loss
return kld_all, ll_all
def train_clvrnn_model(model, sess, config, input_data, target_data,
trace_lengths, selection_matrix):
global TRAINING_ITERATIONS
beta_kld = TRAINING_ITERATIONS * 0.00001 if TRAINING_ITERATIONS * 0.00001 < 1 else 1
[output_x, kl_loss, likelihood_loss, _] = sess.run(
[
model.output, model.kl_loss, model.likelihood_loss,
model.train_general_op
],
feed_dict={
model.input_data_ph: input_data,
model.target_data_ph: target_data,
model.trace_length_ph: trace_lengths,
model.selection_matrix_ph: selection_matrix,
model.kld_beta: beta_kld,
# model.kld_beta: 0,
})
output_decoder = []
prior_output_decoder = []
kld_all = []
ll_all = []
for batch_index in range(0, len(output_x)):
output_decoder_batch = []
prior_output_decoder_batch = []
for trace_length_index in range(0, config.Learn.max_seq_length):
if selection_matrix[batch_index][trace_length_index]:
output_decoder_batch.append(
output_x[batch_index][trace_length_index])
kld_all.append(kl_loss[batch_index][trace_length_index])
ll_all.append(likelihood_loss[batch_index][trace_length_index])
else:
output_decoder_batch.append(
np.asarray([0] * config.Arch.CLVRNN.x_dim))
output_decoder.append(output_decoder_batch)
prior_output_decoder.append(prior_output_decoder_batch)
output_decoder = np.asarray(output_decoder)
acc = compute_rnn_acc(output_prob=output_decoder,
target_label=target_data,
selection_matrix=selection_matrix,
config=config)
if TRAINING_ITERATIONS % 20 == 0:
print(
"Embed acc is {0} with mean KLD:{1} and mean ll:{2} with beta:{3} "
.format(acc, np.mean(kld_all), np.mean(ll_all), beta_kld))
TRAINING_ITERATIONS += 1
# print acc
# if cost_out > 0.0001: # TODO: we still need to consider how to define convergence
# converge_flag = False
cost_out = likelihood_loss + kl_loss
return kld_all, ll_all
def validate_model(testing_dir_games_all,
data_store,
config,
sess,
model,
player_id_cluster_dir,
train_game_number,
validate_cvrnn_flag=True,
validate_td_flag=True):
output_decoder_all = None
target_data_all = None
selection_matrix_all = None
q_values_all = None
validate_variance_flag = False
print('validating model')
for dir_game in testing_dir_games_all:
print('validating game {0}'.format(str(dir_game)))
if dir_game == '.DS_Store':
continue
[
output_decoder_all, target_data_all, selection_matrix_all,
q_values_all, match_q_values_players_dict
] = gathering_running_and_run(
dir_game,
config,
player_id_cluster_dir,
data_store,
model,
sess,
training_flag=False,
game_number=None,
validate_cvrnn_flag=validate_cvrnn_flag,
validate_td_flag=validate_td_flag,
validate_variance_flag=validate_variance_flag,
output_decoder_all=output_decoder_all,
target_data_all=target_data_all,
selection_matrix_all=selection_matrix_all,
q_values_all=q_values_all)
# validate_variance_flag = False
if match_q_values_players_dict is not None:
plot_players_games(match_q_values_players_dict, train_game_number)
if validate_cvrnn_flag:
acc = compute_rnn_acc(output_prob=output_decoder_all,
target_label=target_data_all,
selection_matrix=selection_matrix_all,
config=config,
if_print=True)
print("testing acc is {0}".format(str(acc)))
if validate_td_flag:
print("testing avg qs is {0}".format(str(np.mean(q_values_all,
axis=0))))
def validate_model(testing_dir_games_all,
data_store,
source_data_dir,
config,
sess,
model,
player_id_cluster_dir,
train_game_number,
validate_cvrnn_flag,
validate_td_flag,
validate_diff_flag,
validate_pred_flag,
file_writer=None):
output_decoder_all = None
target_data_all = None
selection_matrix_all = None
q_values_all = None
validate_variance_flag = False
pred_target_data_all = None
pred_output_prob_all = None
if validate_diff_flag:
real_label_record = np.ones([len(testing_dir_games_all), 5000]) * -1
output_label_record = np.ones([len(testing_dir_games_all), 5000]) * -1
print('validating model')
for dir_index in range(0, len(testing_dir_games_all)):
real_label_all = None
output_label_all = None
dir_game = testing_dir_games_all[dir_index]
print('validating game {0}'.format(str(dir_game)))
if dir_game == '.DS_Store':
continue
[
output_decoder_all, target_data_all, selection_matrix_all,
q_values_all, real_label_all, output_label_all,
pred_target_data_all, pred_output_prob_all,
match_q_values_players_dict
] = gathering_data_and_run(
dir_game,
config,
player_id_cluster_dir,
data_store,
source_data_dir,
model,
sess,
training_flag=False,
game_number=None,
validate_cvrnn_flag=validate_cvrnn_flag,
validate_td_flag=validate_td_flag,
validate_diff_flag=validate_diff_flag,
validate_variance_flag=validate_variance_flag,
validate_predict_flag=validate_pred_flag,
output_decoder_all=output_decoder_all,
target_data_all=target_data_all,
selection_matrix_all=selection_matrix_all,
q_values_all=q_values_all,
output_label_all=output_label_all,
real_label_all=real_label_all,
pred_target_data_all=pred_target_data_all,
pred_output_prob_all=pred_output_prob_all)
# validate_variance_flag = False
# if match_q_values_players_dict is not None:
# plot_players_games(match_q_values_players_dict, train_game_number)
if validate_diff_flag:
real_label_record[dir_index][:len(
real_label_all)] = real_label_all[:len(real_label_all)]
output_label_record[dir_index][:len(
output_label_all)] = output_label_all[:len(output_label_all)]
if validate_cvrnn_flag:
acc = compute_rnn_acc(output_prob=output_decoder_all,
target_label=target_data_all,
selection_matrix=selection_matrix_all,
config=config,
if_print=True)
print("testing acc is {0}".format(str(acc)))
if file_writer is not None:
file_writer.write("testing acc is {0}\n".format(str(acc)))
def train_cvrnn_model(model,
sess,
config,
input_data,
target_data,
trace_lengths,
selection_matrix,
pretrain_flag=False):
if pretrain_flag:
[
output_x,
# total_loss,
kl_loss,
likelihood_loss,
_
] = sess.run(
[
model.output,
# model.cost,
model.kl_loss,
model.likelihood_loss,
model.train_ll_op
],
feed_dict={
model.input_data_ph: input_data,
model.target_data_ph: target_data,
model.trace_length_ph: trace_lengths,
model.selection_matrix_ph: selection_matrix
})
else:
[
output_x,
# total_loss,
kl_loss,
likelihood_loss,
_
] = sess.run(
[
model.output,
# model.cost,
model.kl_loss,
model.likelihood_loss,
model.train_general_op
],
feed_dict={
model.input_data_ph: input_data,
model.target_data_ph: target_data,
model.trace_length_ph: trace_lengths,
model.selection_matrix_ph: selection_matrix
})
output_decoder = []
for batch_index in range(0, len(output_x)):
output_decoder_batch = []
for trace_length_index in range(0, config.Learn.max_seq_length):
if selection_matrix[batch_index][trace_length_index]:
output_decoder_batch.append(
output_x[batch_index][trace_length_index])
else:
output_decoder_batch.append(
np.asarray([0] * config.Arch.CVRNN.x_dim))
output_decoder.append(output_decoder_batch)
output_decoder = np.asarray(output_decoder)
acc = compute_rnn_acc(output_prob=output_decoder,
target_label=target_data,
selection_matrix=selection_matrix,
config=config)
# print acc
# if cost_out > 0.0001: # TODO: we still need to consider how to define convergence
# converge_flag = False
cost_out = likelihood_loss + kl_loss
def validate_model(testing_dir_games_all,
data_store,
source_data_dir,
config,
sess,
model,
player_id_cluster_dir,
train_game_number,
validate_cvrnn_flag,
validate_td_flag,
validate_diff_flag,
validate_pred_flag,
file_writer=None):
output_decoder_all = None
target_data_all = None
selection_matrix_all = None
q_values_all = None
validate_variance_flag = False
pred_target_data_all = None
pred_output_prob_all = None
if validate_diff_flag:
real_label_record = np.ones([len(testing_dir_games_all), 5000]) * -1
output_label_record = np.ones([len(testing_dir_games_all), 5000]) * -1
print('validating model')
for dir_index in range(0, len(testing_dir_games_all)):
real_label_all = None
output_label_all = None
dir_game = testing_dir_games_all[dir_index]
print('validating game {0}'.format(str(dir_game)))
if dir_game == '.DS_Store':
continue
[
output_decoder_all, target_data_all, selection_matrix_all,
q_values_all, real_label_all, output_label_all,
pred_target_data_all, pred_output_prob_all,
match_q_values_players_dict
] = gathering_data_and_run(
dir_game,
config,
player_id_cluster_dir,
data_store,
source_data_dir,
model,
sess,
training_flag=False,
game_number=None,
validate_cvrnn_flag=validate_cvrnn_flag,
validate_td_flag=validate_td_flag,
validate_diff_flag=validate_diff_flag,
validate_variance_flag=validate_variance_flag,
validate_predict_flag=validate_pred_flag,
output_decoder_all=output_decoder_all,
target_data_all=target_data_all,
selection_matrix_all=selection_matrix_all,
q_values_all=q_values_all,
output_label_all=output_label_all,
real_label_all=real_label_all,
pred_target_data_all=pred_target_data_all,
pred_output_prob_all=pred_output_prob_all)
# validate_variance_flag = False
# if match_q_values_players_dict is not None:
# plot_players_games(match_q_values_players_dict, train_game_number)
if validate_diff_flag:
real_label_record[dir_index][:len(
real_label_all)] = real_label_all[:len(real_label_all)]
output_label_record[dir_index][:len(
output_label_all)] = output_label_all[:len(output_label_all)]
if validate_cvrnn_flag:
acc = compute_rnn_acc(output_prob=output_decoder_all,
target_label=target_data_all,
selection_matrix=selection_matrix_all,
config=config,
if_print=True)
print("testing acc is {0}".format(str(acc)))
if file_writer is not None:
file_writer.write("testing acc is {0}\n".format(str(acc)))
# if validate_td_flag:
# print ("testing avg qs is {0}".format(str(np.mean(q_values_all, axis=0))))
# if file_writer is not None:
# file_writer.write("testing avg qs is {0}\n".format(str(np.mean(q_values_all, axis=0))))
if validate_diff_flag:
# print ('general real label is {0}'.format(str(np.sum(real_label_record, axis=1))))
# print ('general output label is {0}'.format(str(np.sum(output_label_record, axis=1))))
for i in range(0, output_label_record.shape[1]):
real_outcome_record_step = real_label_record[:, i]
model_output_record_step = output_label_record[:, i]
diff_sum = 0
total_number = 0
print_flag = True
for win_index in range(0, len(real_outcome_record_step)):
if model_output_record_step[
win_index] == -100 or real_outcome_record_step[
win_index] == -100:
print_flag = True
continue
diff = abs(model_output_record_step[win_index] -
real_outcome_record_step[win_index])
diff_sum += diff
total_number += 1
if print_flag:
if i % 100 == 0 and total_number > 0:
print('diff of time {0} is {1}'.format(
str(i), str(float(diff_sum) / total_number)))
if file_writer is not None:
file_writer.write('diff of time {0} is {1}\n'.format(
str(i), str(float(diff_sum) / total_number)))
if validate_pred_flag:
TP, TN, FP, FN, acc, ll, auc = compute_acc(pred_target_data_all,
pred_output_prob_all,
if_binary_result=True,
if_print=False,
if_add_ll=True)
precision = float(TP) / (TP + FP)
recall = float(TP) / (TP + FN)
print("Prediction acc is {0} with precision {1} and recall {2}".format(
acc, precision, recall))