1.CNN卷积神经网络
(1)代码
import torch
import torch.nn as nn
import torch.utils.data as Data
import torchvision # 数据库模块
import matplotlib.pyplot as plt
torch.manual_seed(1) # reproducible
# Hyper Parameters
EPOCH = 1 # 训练整批数据多少次, 为了节约时间, 我们只训练一次
BATCH_SIZE = 50
LR = 0.001 # 学习率
DOWNLOAD_MNIST = True # 如果你已经下载好了mnist数据就写上 False
# Mnist 手写数字
train_data = torchvision.datasets.MNIST(
root='./mnist/', # 保存或者提取位置
train=True, # this is training data
transform=torchvision.transforms.ToTensor(), # 转换 PIL.Image or numpy.ndarray 成
# torch.FloatTensor (C x H x W), 训练的时候 normalize 成 [0.0, 1.0] 区间
download=DOWNLOAD_MNIST, # 没下载就下载, 下载了就不用再下了
)
#plot one example
print(train_data.train_data.size)
print(train_data.train_labels.size())
plt.imshow(train_data.train_data[0].numpy(),cmap='gray')
plt.title('%i'%train_data.train_labels[0])
plt.show()
(2)运行结果
torch.Size([60000])
Process finished with exit code 0
2.打印CNN
(1)代码
import torch
import torch.nn as nn
import torch.utils.data as Data
import torchvision # 数据库模块
import matplotlib.pyplot as plt
from torch.autograd import Variable
torch.manual_seed(1) # reproducible
# Hyper Parameters
EPOCH = 1 # 训练整批数据多少次, 为了节约时间, 我们只训练一次
BATCH_SIZE = 50
LR = 0.001 # 学习率
DOWNLOAD_MNIST = False # 如果你已经下载好了mnist数据就写上 False
# Mnist 手写数字
train_data = torchvision.datasets.MNIST(
root='./mnist/', # 保存或者提取位置
train=True, # this is training data
transform=torchvision.transforms.ToTensor(), # 转换 PIL.Image or numpy.ndarray 成
# torch.FloatTensor (C x H x W), 训练的时候 normalize 成 [0.0, 1.0] 区间
download=DOWNLOAD_MNIST, # 没下载就下载, 下载了就不用再下了
)
#plot one example
# print(train_data.train_data.size)
# print(train_data.train_labels.size())
# plt.imshow(train_data.train_data[0].numpy(),cmap='gray')
# plt.title('%i'%train_data.train_labels[0])
# plt.show()
train_loader = Data.DataLoader(dataset=train_data,batch_size=BATCH_SIZE,shuffle=True,num_workers=2)
test_data = torchvision.datasets.MNIST(root='./mnist/', train=False)#意味着不是train_data,而是testdata
# # 批训练 50samples, 1 channel, 28x28 (50, 1, 28, 28)
# train_loader = Data.DataLoader(dataset=train_data, batch_size=BATCH_SIZE, shuffle=True)
# # 为了节约时间, 测试时只测试前2000个
test_x = torch.unsqueeze(test_data.test_data, dim=1).type(torch.FloatTensor)[:2000]/255. # shape from (2000, 28, 28) to (2000, 1, 28, 28), value in range(0,1)
# test_x = Variable(torch.unsqueeze(test_data.test_data,dim=1),volatile=True).type(torch.FloatTensor)[:2000]/255.#运行不通
test_y = test_data.test_labels[:2000]
# 建立CNN神经网络
class CNN(nn.Module):
def __init__(self):
super(CNN,self).__init__()
self.conv1 = nn.Sequential(
nn.Conv2d( #(1,28,28)
in_channels=1,#输入图片层灰度图1层,RGB3
out_channels=16,#16个特征,即高度16
kernel_size=5,#filter 长宽为5个像素点
stride=1,#每隔多少像素跳动一下
padding=2,#原始图片周围围上0,为了不丢失边缘数据
#if strid = 1,padding = (kernel_size-1)/2 = (5-1)/2=2
# wn+1=(wn+p*2-k)/s+1
),#卷积层——过滤器 filter ;(16,28,28)
nn.ReLU(),#神经网络--非线性激活层
nn.MaxPool2d(kernel_size=2),#池化层,若不池化,数据大,需stride>1;(16,14,14)
)
self.conv2 = nn.Sequential(
nn.Conv2d(16,32,5,1,2),#(32,14,14)32= 28+2*2-5/1+1
nn.ReLU(),#(32,14,14)
nn.MaxPool2d(2) #(32,7,7)
)
self.out = nn.Linear(32*7*7,10)
def forward(self,x):
x = self.conv1(x)
x = self.conv2(x)#batch,32,7,7
x = x.view(x.size(0),-1) #batch,32*7*7
output=self.out(x)
return output
cnn = CNN()
print(cnn)
(2)运行结果
CNN(
(conv1): Sequential(
(0): Conv2d(1, 16, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
(1): ReLU()
(2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
)
(conv2): Sequential(
(0): Conv2d(16, 32, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
(1): ReLU()
(2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
)
(out): Linear(in_features=1568, out_features=10, bias=True)
)
Process finished with exit code 0
3.训练CNN
(1)代码
import torch
import torch.nn as nn
import torch.utils.data as Data
import torchvision # 数据库模块
import matplotlib.pyplot as plt
from torch.autograd import Variable
torch.manual_seed(1) # reproducible
# Hyper Parameters
EPOCH = 1 # 训练整批数据多少次, 为了节约时间, 我们只训练一次
BATCH_SIZE = 50 # 批训练的数据个数
LR = 0.001 # 学习率
DOWNLOAD_MNIST = False # 如果你已经下载好了mnist数据就写上 False
# Mnist 手写数字
train_data = torchvision.datasets.MNIST(
root='./mnist/', # 保存或者提取位置
train=True, # this is training data
transform=torchvision.transforms.ToTensor(), # 转换 PIL.Image or numpy.ndarray 成
# torch.FloatTensor (C x H x W), 训练的时候 normalize 成 [0.0, 1.0] 区间
download=DOWNLOAD_MNIST, # 没下载就下载, 下载了就不用再下了
)
#plot one example
# print(train_data.train_data.size)
# print(train_data.train_labels.size())
# plt.imshow(train_data.train_data[0].numpy(),cmap='gray')
# plt.title('%i'%train_data.train_labels[0])
# plt.show()
train_loader = Data.DataLoader(dataset=train_data,batch_size=BATCH_SIZE,shuffle=True)## num_workers=2,多线程来读数据 windows删除此行
test_data = torchvision.datasets.MNIST(root='./mnist/', train=False)#意味着不是train_data,而是testdata
# # 批训练 50samples, 1 channel, 28x28 (50, 1, 28, 28)
# train_loader = Data.DataLoader(dataset=train_data, batch_size=BATCH_SIZE, shuffle=True)
# # 为节约时间, 测试前2000个
test_x = torch.unsqueeze(test_data.test_data, dim=1).type(torch.FloatTensor)[:2000]/255. # shape from (2000, 28, 28) to (2000, 1, 28, 28), value in range(0,1)
test_y = test_data.test_labels[:2000]
# 建立CNN神经网络
class CNN(nn.Module):
def __init__(self):
super(CNN,self).__init__()
self.conv1 = nn.Sequential(
nn.Conv2d( #(1,28,28)
in_channels=1,#输入图片层灰度图1层,RGB3
out_channels=16,#16个特征,即高度16 filter
kernel_size=5,#filter size 长宽为5个像素点
stride=1,#每隔多少像素跳动一下
padding=2,#原始图片周围围上0,为了不丢失边缘数据
#if strid = 1,padding = (kernel_size-1)/2 = (5-1)/2=2
# wn+1=(wn+p*2-k)/s+1
),#卷积层——过滤器 filter ;(16,28,28)
nn.ReLU(),#神经网络--非线性激活层
nn.MaxPool2d(kernel_size=2),#池化层,若不池化,数据大,需stride>1;(16,14,14)
)
self.conv2 = nn.Sequential(
nn.Conv2d(16,32,5,1,2),#(32,14,14)32= 28+2*2-5/1+1
nn.ReLU(),#(32,14,14)
nn.MaxPool2d(2) #(32,7,7)
)
self.out = nn.Linear(32*7*7,10)
def forward(self,x):
x = self.conv1(x)
x = self.conv2(x)#batch,32,7,7
x = x.view(x.size(0),-1) #batch,32*7*7
output=self.out(x)
return output,x
cnn = CNN()
print(cnn)
optimizer = torch.optim.Adam(cnn.parameters(), lr=LR) # optimize all cnn parameters
loss_func = nn.CrossEntropyLoss()
# following function (plot_with_labels) is for visualization, can be ignored if not interested
from matplotlib import cm
try: from sklearn.manifold import TSNE; HAS_SK = True
except: HAS_SK = False; print('Please install sklearn for layer visualization')
def plot_with_labels(lowDWeights, labels):
plt.cla()
X, Y = lowDWeights[:, 0], lowDWeights[:, 1]
for x, y, s in zip(X, Y, labels):
c = cm.rainbow(int(255 * s / 9)); plt.text(x, y, s, backgroundcolor=c, fontsize=9)
plt.xlim(X.min(), X.max()); plt.ylim(Y.min(), Y.max()); plt.title('Visualize last layer'); plt.show(); plt.pause(0.01)
plt.ion()
# training and testing
for epoch in range(EPOCH):
for step, (b_x, b_y) in enumerate(train_loader): # gives batch data, normalize x when iterate train_loader
output = cnn(b_x)[0] # cnn output
loss = loss_func(output, b_y) # cross entropy loss
optimizer.zero_grad() # clear gradients for this training step
loss.backward() # backpropagation, compute gradients
optimizer.step() # apply gradients
if step % 50 == 0:
test_output, last_layer = cnn(test_x)
pred_y = torch.max(test_output, 1)[1].data.numpy()
accuracy = float((pred_y == test_y.data.numpy()).astype(int).sum()) / float(test_y.size(0))
print('Epoch: ', epoch, '| train loss: %.4f' % loss.data.numpy(), '| test accuracy: %.2f' % accuracy)
if HAS_SK:
# Visualization of trained flatten layer (T-SNE)
tsne = TSNE(perplexity=30, n_components=2, init='pca', n_iter=5000)
plot_only = 500
low_dim_embs = tsne.fit_transform(last_layer.data.numpy()[:plot_only, :])
labels = test_y.numpy()[:plot_only]
plot_with_labels(low_dim_embs, labels)
plt.ioff()
# print 10 predictions from test data
test_output, _ = cnn(test_x[:10])
pred_y = torch.max(test_output, 1)[1].data.numpy()
print(pred_y, 'prediction number')
print(test_y[:10].numpy(), 'real number')
(2)运行结果
CNN(
(conv1): Sequential(
(0): Conv2d(1, 16, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
(1): ReLU()
(2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
)
(conv2): Sequential(
(0): Conv2d(16, 32, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
(1): ReLU()
(2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
)
(out): Linear(in_features=1568, out_features=10, bias=True)
)
Epoch: 0 | train loss: 2.3105 | test accuracy: 0.06
Epoch: 0 | train loss: 0.6184 | test accuracy: 0.83
Epoch: 0 | train loss: 0.1290 | test accuracy: 0.87
Epoch: 0 | train loss: 0.2371 | test accuracy: 0.91
Epoch: 0 | train loss: 0.4058 | test accuracy: 0.93
Epoch: 0 | train loss: 0.0850 | test accuracy: 0.94
Epoch: 0 | train loss: 0.1956 | test accuracy: 0.94
Epoch: 0 | train loss: 0.1087 | test accuracy: 0.95
Epoch: 0 | train loss: 0.1238 | test accuracy: 0.96
Epoch: 0 | train loss: 0.0703 | test accuracy: 0.96
Epoch: 0 | train loss: 0.2217 | test accuracy: 0.96
Epoch: 0 | train loss: 0.2101 | test accuracy: 0.96
Epoch: 0 | train loss: 0.0237 | test accuracy: 0.97
Epoch: 0 | train loss: 0.0844 | test accuracy: 0.97
Epoch: 0 | train loss: 0.2158 | test accuracy: 0.97
Epoch: 0 | train loss: 0.1007 | test accuracy: 0.97
Epoch: 0 | train loss: 0.0433 | test accuracy: 0.97
Epoch: 0 | train loss: 0.0923 | test accuracy: 0.98
Epoch: 0 | train loss: 0.0564 | test accuracy: 0.98
Epoch: 0 | train loss: 0.1006 | test accuracy: 0.98
Epoch: 0 | train loss: 0.0320 | test accuracy: 0.98
Epoch: 0 | train loss: 0.0229 | test accuracy: 0.98
Epoch: 0 | train loss: 0.0233 | test accuracy: 0.98
Epoch: 0 | train loss: 0.1208 | test accuracy: 0.98
[7 2 1 0 4 1 4 9 5 9] prediction number
[7 2 1 0 4 1 4 9 5 9] real number
Process finished with exit code 0