def forward(self, seq_dignosis_codes, seq_time_step, batch_labels, options,
maxlen):
seq_time_step = np.array(list(units.pad_time(seq_time_step, options)))
lengths = torch.from_numpy(
np.array([len(seq) for seq in seq_dignosis_codes])).cuda()
diagnosis_codes, labels, mask, mask_final, mask_code = units.pad_matrix_new(
seq_dignosis_codes, batch_labels, options)
if options['use_gpu']:
diagnosis_codes = torch.LongTensor(diagnosis_codes).cuda()
mask_mult = torch.BoolTensor(1 - mask).unsqueeze(2).cuda()
mask_final = torch.Tensor(mask_final).unsqueeze(2).cuda()
mask_code = torch.Tensor(mask_code).unsqueeze(3).cuda()
else:
diagnosis_codes = torch.LongTensor(diagnosis_codes)
mask_mult = torch.BoolTensor(1 - mask).unsqueeze(2)
mask_final = torch.Tensor(mask_final).unsqueeze(2)
mask_code = torch.Tensor(mask_code).unsqueeze(3)
features = self.feature_encoder(diagnosis_codes, mask_mult, mask_code,
seq_time_step, lengths)
final_statues = features * mask_final
final_statues = final_statues.sum(1)
predictions = self.classify_layer(final_statues)
labels = torch.LongTensor(labels)
if options['use_gpu']:
labels = labels.cuda()
return predictions, labels, None
def forward(self, seq_dignosis_codes, seq_time_step, batch_labels, options,
maxlen):
# seq_dignosis_codes: [batch_size, length, bag_len]
# seq_time_step: [batch_size, length] the day times to the final visit
# batch_labels: [batch_size] 0 negative 1 positive
seq_time_step = np.array(list(units.pad_time(seq_time_step, options)))
lengths = torch.from_numpy(
np.array([len(seq) for seq in seq_dignosis_codes])).cuda()
diagnosis_codes, labels, mask, mask_final, mask_code = units.pad_matrix_new(
seq_dignosis_codes, batch_labels, options)
if options['use_gpu']:
diagnosis_codes = torch.LongTensor(diagnosis_codes).cuda()
mask_mult = torch.BoolTensor(1 - mask).unsqueeze(2).cuda()
mask_final = torch.Tensor(mask_final).unsqueeze(2).cuda()
mask_code = torch.Tensor(mask_code).unsqueeze(3).cuda()
else:
diagnosis_codes = torch.LongTensor(diagnosis_codes)
mask_mult = torch.BoolTensor(1 - mask).unsqueeze(2)
mask_final = torch.Tensor(mask_final).unsqueeze(2)
mask_code = torch.Tensor(mask_code).unsqueeze(3)
features = self.feature_encoder(diagnosis_codes, mask_mult, mask_code,
seq_time_step, lengths)
final_statues = features * mask_final
final_statues = final_statues.sum(1, keepdim=True)
quiryes = self.relu(self.quiry_layer(final_statues))
# prior_weight = self.prior_encoder(seq_dignosis_codes, maxlen, quiryes, options, mask_mult)
self_weight = self.get_self_attention(features, quiryes, mask_mult)
time_weight = self.time_encoder(seq_time_step, quiryes, options,
mask_mult)
attention_weight = torch.softmax(
self.quiry_weight_layer(final_statues), 2)
total_weight = torch.cat((time_weight, self_weight), 2)
total_weight = torch.sum(total_weight * attention_weight,
2,
keepdim=True)
total_weight = total_weight / (
torch.sum(total_weight, 1, keepdim=True) + 1e-5)
weighted_features = features * total_weight
averaged_features = torch.sum(weighted_features, 1)
averaged_features = self.dropout(averaged_features)
predictions = self.classify_layer(averaged_features)
labels = torch.LongTensor(labels)
if options['use_gpu']:
labels = labels.cuda()
return predictions, labels, self_weight