北京网站建设价格便宜,建设网站烧钱,大学生网站作业,手机怎么制作图片关于
大规模数据收集和注释的成本通常使得将机器学习算法应用于新任务或数据集变得异常昂贵。规避这一成本的一种方法是在合成数据上训练模型#xff0c;其中自动提供注释。尽管它们很有吸引力#xff0c;但此类模型通常无法从合成图像推广到真实图像#xff0c;因此需要域…关于
大规模数据收集和注释的成本通常使得将机器学习算法应用于新任务或数据集变得异常昂贵。规避这一成本的一种方法是在合成数据上训练模型其中自动提供注释。尽管它们很有吸引力但此类模型通常无法从合成图像推广到真实图像因此需要域适应算法来操纵这些模型然后才能成功应用。现有的方法要么侧重于将表示从一个域映射到另一个域要么侧重于学习提取对于提取它们的域而言不变的特征。然而通过只关注在两个域之间创建映射或共享表示他们忽略了每个域的单独特征。域分离网络可以实现对每个域的独特之处进行特征建模,同时进行模型域不变特征的提取。 参考文章 https://arxiv.org/abs/1608.06019
工具 方法实现
数据集定义
import torch.utils.data as data
from PIL import Image
import osclass GetLoader(data.Dataset):def __init__(self, data_root, data_list, transformNone):self.root data_rootself.transform transformf open(data_list, r)data_list f.readlines()f.close()self.n_data len(data_list)self.img_paths []self.img_labels []for data in data_list:self.img_paths.append(data[:-3])self.img_labels.append(data[-2])def __getitem__(self, item):img_paths, labels self.img_paths[item], self.img_labels[item]imgs Image.open(os.path.join(self.root, img_paths)).convert(RGB)if self.transform is not None:imgs self.transform(imgs)labels int(labels)return imgs, labelsdef __len__(self):return self.n_data
模型搭建
import torch.nn as nn
from functions import ReverseLayerFclass DSN(nn.Module):def __init__(self, code_size100, n_class10):super(DSN, self).__init__()self.code_size code_size########################################### private source encoder##########################################self.source_encoder_conv nn.Sequential()self.source_encoder_conv.add_module(conv_pse1, nn.Conv2d(in_channels3, out_channels32, kernel_size5,padding2))self.source_encoder_conv.add_module(ac_pse1, nn.ReLU(True))self.source_encoder_conv.add_module(pool_pse1, nn.MaxPool2d(kernel_size2, stride2))self.source_encoder_conv.add_module(conv_pse2, nn.Conv2d(in_channels32, out_channels64, kernel_size5,padding2))self.source_encoder_conv.add_module(ac_pse2, nn.ReLU(True))self.source_encoder_conv.add_module(pool_pse2, nn.MaxPool2d(kernel_size2, stride2))self.source_encoder_fc nn.Sequential()self.source_encoder_fc.add_module(fc_pse3, nn.Linear(in_features7 * 7 * 64, out_featurescode_size))self.source_encoder_fc.add_module(ac_pse3, nn.ReLU(True))########################################## private target encoder#########################################self.target_encoder_conv nn.Sequential()self.target_encoder_conv.add_module(conv_pte1, nn.Conv2d(in_channels3, out_channels32, kernel_size5,padding2))self.target_encoder_conv.add_module(ac_pte1, nn.ReLU(True))self.target_encoder_conv.add_module(pool_pte1, nn.MaxPool2d(kernel_size2, stride2))self.target_encoder_conv.add_module(conv_pte2, nn.Conv2d(in_channels32, out_channels64, kernel_size5,padding2))self.target_encoder_conv.add_module(ac_pte2, nn.ReLU(True))self.target_encoder_conv.add_module(pool_pte2, nn.MaxPool2d(kernel_size2, stride2))self.target_encoder_fc nn.Sequential()self.target_encoder_fc.add_module(fc_pte3, nn.Linear(in_features7 * 7 * 64, out_featurescode_size))self.target_encoder_fc.add_module(ac_pte3, nn.ReLU(True))################################# shared encoder (dann_mnist)################################self.shared_encoder_conv nn.Sequential()self.shared_encoder_conv.add_module(conv_se1, nn.Conv2d(in_channels3, out_channels32, kernel_size5,padding2))self.shared_encoder_conv.add_module(ac_se1, nn.ReLU(True))self.shared_encoder_conv.add_module(pool_se1, nn.MaxPool2d(kernel_size2, stride2))self.shared_encoder_conv.add_module(conv_se2, nn.Conv2d(in_channels32, out_channels48, kernel_size5,padding2))self.shared_encoder_conv.add_module(ac_se2, nn.ReLU(True))self.shared_encoder_conv.add_module(pool_se2, nn.MaxPool2d(kernel_size2, stride2))self.shared_encoder_fc nn.Sequential()self.shared_encoder_fc.add_module(fc_se3, nn.Linear(in_features7 * 7 * 48, out_featurescode_size))self.shared_encoder_fc.add_module(ac_se3, nn.ReLU(True))# classify 10 numbersself.shared_encoder_pred_class nn.Sequential()self.shared_encoder_pred_class.add_module(fc_se4, nn.Linear(in_featurescode_size, out_features100))self.shared_encoder_pred_class.add_module(relu_se4, nn.ReLU(True))self.shared_encoder_pred_class.add_module(fc_se5, nn.Linear(in_features100, out_featuresn_class))self.shared_encoder_pred_domain nn.Sequential()self.shared_encoder_pred_domain.add_module(fc_se6, nn.Linear(in_features100, out_features100))self.shared_encoder_pred_domain.add_module(relu_se6, nn.ReLU(True))# classify two domainself.shared_encoder_pred_domain.add_module(fc_se7, nn.Linear(in_features100, out_features2))####################################### shared decoder (small decoder)######################################self.shared_decoder_fc nn.Sequential()self.shared_decoder_fc.add_module(fc_sd1, nn.Linear(in_featurescode_size, out_features588))self.shared_decoder_fc.add_module(relu_sd1, nn.ReLU(True))self.shared_decoder_conv nn.Sequential()self.shared_decoder_conv.add_module(conv_sd2, nn.Conv2d(in_channels3, out_channels16, kernel_size5,padding2))self.shared_decoder_conv.add_module(relu_sd2, nn.ReLU())self.shared_decoder_conv.add_module(conv_sd3, nn.Conv2d(in_channels16, out_channels16, kernel_size5,padding2))self.shared_decoder_conv.add_module(relu_sd3, nn.ReLU())self.shared_decoder_conv.add_module(us_sd4, nn.Upsample(scale_factor2))self.shared_decoder_conv.add_module(conv_sd5, nn.Conv2d(in_channels16, out_channels16, kernel_size3,padding1))self.shared_decoder_conv.add_module(relu_sd5, nn.ReLU(True))self.shared_decoder_conv.add_module(conv_sd6, nn.Conv2d(in_channels16, out_channels3, kernel_size3,padding1))def forward(self, input_data, mode, rec_scheme, p0.0):result []if mode source:# source private encoderprivate_feat self.source_encoder_conv(input_data)private_feat private_feat.view(-1, 64 * 7 * 7)private_code self.source_encoder_fc(private_feat)elif mode target:# target private encoderprivate_feat self.target_encoder_conv(input_data)private_feat private_feat.view(-1, 64 * 7 * 7)private_code self.target_encoder_fc(private_feat)result.append(private_code)# shared encodershared_feat self.shared_encoder_conv(input_data)shared_feat shared_feat.view(-1, 48 * 7 * 7)shared_code self.shared_encoder_fc(shared_feat)result.append(shared_code)reversed_shared_code ReverseLayerF.apply(shared_code, p)domain_label self.shared_encoder_pred_domain(reversed_shared_code)result.append(domain_label)if mode source:class_label self.shared_encoder_pred_class(shared_code)result.append(class_label)# shared decoderif rec_scheme share:union_code shared_codeelif rec_scheme all:union_code private_code shared_codeelif rec_scheme private:union_code private_coderec_vec self.shared_decoder_fc(union_code)rec_vec rec_vec.view(-1, 3, 14, 14)rec_code self.shared_decoder_conv(rec_vec)result.append(rec_code)return result 模型训练
import random
import os
import torch.backends.cudnn as cudnn
import torch.optim as optim
import torch.utils.data
import numpy as np
from torch.autograd import Variable
from torchvision import datasets
from torchvision import transforms
from model_compat import DSN
from data_loader import GetLoader
from functions import SIMSE, DiffLoss, MSE
from test import test######################
# params #
######################source_image_root os.path.join(., dataset, mnist)
target_image_root os.path.join(., dataset, mnist_m)
model_root model
cuda True
cudnn.benchmark True
lr 1e-2
batch_size 32
image_size 28
n_epoch 100
step_decay_weight 0.95
lr_decay_step 20000
active_domain_loss_step 10000
weight_decay 1e-6
alpha_weight 0.01
beta_weight 0.075
gamma_weight 0.25
momentum 0.9manual_seed random.randint(1, 10000)
random.seed(manual_seed)
torch.manual_seed(manual_seed)#######################
# load data #
#######################img_transform transforms.Compose([transforms.Resize(image_size),transforms.ToTensor(),transforms.Normalize(mean(0.5, 0.5, 0.5), std(0.5, 0.5, 0.5))
])dataset_source datasets.MNIST(rootsource_image_root,trainTrue,transformimg_transform
)dataloader_source torch.utils.data.DataLoader(datasetdataset_source,batch_sizebatch_size,shuffleTrue,num_workers8
)train_list os.path.join(target_image_root, mnist_m_train_labels.txt)dataset_target GetLoader(data_rootos.path.join(target_image_root, mnist_m_train),data_listtrain_list,transformimg_transform
)dataloader_target torch.utils.data.DataLoader(datasetdataset_target,batch_sizebatch_size,shuffleTrue,num_workers8
)#####################
# load model #
#####################my_net DSN()#####################
# setup optimizer #
#####################def exp_lr_scheduler(optimizer, step, init_lrlr, lr_decay_steplr_decay_step, step_decay_weightstep_decay_weight):# Decay learning rate by a factor of step_decay_weight every lr_decay_stepcurrent_lr init_lr * (step_decay_weight ** (step / lr_decay_step))if step % lr_decay_step 0:print learning rate is set to %f % current_lrfor param_group in optimizer.param_groups:param_group[lr] current_lrreturn optimizeroptimizer optim.SGD(my_net.parameters(), lrlr, momentummomentum, weight_decayweight_decay)loss_classification torch.nn.CrossEntropyLoss()
loss_recon1 MSE()
loss_recon2 SIMSE()
loss_diff DiffLoss()
loss_similarity torch.nn.CrossEntropyLoss()if cuda:my_net my_net.cuda()loss_classification loss_classification.cuda()loss_recon1 loss_recon1.cuda()loss_recon2 loss_recon2.cuda()loss_diff loss_diff.cuda()loss_similarity loss_similarity.cuda()for p in my_net.parameters():p.requires_grad True#############################
# training network #
############################# MNIST数据重建/共有部分特征/私有数据特征可视化 MNIST_m数据重建/共有部分特征/私有数据特征可视化
代码获取
相关问题和项目开发欢迎私信交流和沟通。
