dim_y = 2
train_size = train_data.shape[0]
test_size = test_data.shape[0]
print("dim_x: ", dim_x, "\ntrain_size:", train_size, "\ntest_size:", test_size)
train_data = np.reshape(train_data, [train_size, dim_x, 1, 1])
test_data = np.reshape(test_data, [test_size, dim_x, 1, 1])
assert train_size % batch_size == 0, '#TrainSize % #BatchSize != 0'
assert test_size % batch_size == 0, '#TestSize % #BatchSize != 0'
train_batch_num = int(train_size / batch_size)
test_batch_num = int(test_size / batch_size)
data_placeholder = tf.placeholder(tf.float32, [None, dim_x, 1, 1])
labels_placeholder = tf.placeholder(tf.float32, [None, dim_y])
result = lenet5(data_placeholder, labels_placeholder, dim_y)
pred_logit = result.softmax.softmax_activation()
accuracy_tensor = result.softmax.evaluate_classifier(labels_placeholder,
phase=pt.Phase.test)
precision_tensor, recall_tensor = result.softmax.evaluate_precision_recall(
labels_placeholder, phase=pt.Phase.test)
_, auroc_tensor = tf.metrics.auc(labels_placeholder, pred_logit)
_, aupr_tensor = tf.metrics.auc(labels_placeholder, pred_logit, curve="PR")
optimizer = tf.train.GradientDescentOptimizer(learning_rate)
train_op = pt.apply_optimizer(optimizer, losses=[result.loss])
#save_path = '/data/cdy/ykq/checkpoints/model_conv2d_{}-{}.cpkt'.format(
# learning_rate, time.strftime("%m-%d-%H%M%S", time.localtime()))
#print("model has been saved: " + save_path)
#runner = pt.train.Runner(save_path)
clatent_dim = head_context_trans.shape[-1]
print("clatent_dim: ", clatent_dim)
input_data = tf.reshape(
tf.concat([
head_trans, head_hierarchy_placeholder, head_context_trans, tail_trans,
tail_hierarchy_placeholder, tail_context_trans
], 1),
[batch_size, latent_dim * 2 + hierarchy_dim * 2 + clatent_dim * 2, 1, 1])
# input_data = tf.reshape(tf.concat([head_trans, head_context_trans,
# tail_trans, tail_context_trans], 1),
# [batch_size, latent_dim*2+clatent_dim*2, 1, 1])
print("input_data shape: ", input_data.shape)
result = lenet5(input_data, labels_placeholder, dim_y)
# print("@@@@@result:", result.softmax.shape)
pred_logit = result.softmax.softmax_activation()
accuracy_tensor = result.softmax.evaluate_classifier(labels_placeholder,
phase=pt.Phase.test)
precision_tensor, recall_tensor = result.softmax.evaluate_precision_recall(
labels_placeholder, phase=pt.Phase.test)
_, auroc_tensor = tf.metrics.auc(labels_placeholder, pred_logit)
_, aupr_tensor = tf.metrics.auc(labels_placeholder, pred_logit, curve="PR")
optimizer = tf.train.GradientDescentOptimizer(learning_rate)
train_op = pt.apply_optimizer(optimizer, losses=[result.loss])
runner = pt.train.Runner()
best_f1 = 0