HackMD- ncku-yee·Last edited by ncku-yee on Nov 2, 2021Linked with GitHubContributed by
PyTorch - CNN 卷積神經網絡 - MNIST手寫數字辨識
1. import 需要的 packages
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable
from torchsummary import summary
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
from keras.datasets import mnist
2. 確認一下使用 cuda 或是 cpu
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
print(device)
3. 資料預處理 (Data Preprcessing)
# 使用 keras 直接載入 MNIST dataset
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()
# Normalization
X_train = X_train.astype('float32') / 255
X_test = X_test.astype('float32') / 255
# 利用 sklearn，將每一類都以 8:2 的比例分成訓練資料和測試資料
features_train, features_test, targets_train, targets_test = train_test_split(X_train, Y_train, test_size = 0.2, random_state = 42)
# 將切好的 data 轉成 tensor 形式
# Training Datasets
featuresTrain = torch.from_numpy(features_train)
targetsTrain = torch.from_numpy(targets_train).type(torch.LongTensor) # data type is long
# Testing Datasets
featuresTest = torch.from_numpy(features_test)
targetsTest = torch.from_numpy(targets_test).type(torch.LongTensor) # data type is long
# 使用 torch.utils.data.TensorDataset 將 train 和 test datasets 存成 tensor 形式
# Pytorch train and test TensorDataset
train = torch.utils.data.TensorDataset(featuresTrain, targetsTrain)
test = torch.utils.data.TensorDataset(featuresTest, targetsTest)
4.超參數設定
LR = 0.01 # Learning Rate
batch_size = 100 # Batch size
n_iters = 10000 # Iterations each epoch
num_epochs = n_iters / (len(features_train) / batch_size)
num_epochs = int(num_epochs) # Epochs
5. 將 Training data 和 Testing data 傳至 dataloader
# torch.utils.data.TensorDatasetDataLoader(dataset, batch_size=1, shuffle=False,...) 為數據加載器
# 組合數據集和採樣器，並在數據集上提供單進程或多進程迭代器
# Pytorch DataLoader
train_loader = torch.utils.data.DataLoader(train, batch_size = batch_size, shuffle = True)
test_loader = torch.utils.data.DataLoader(test, batch_size = batch_size, shuffle = True)
6. 建立 CNN 模型
建立 CNN model ，需要注意的是 input shape 在經過每一層後的變化，在設計的時候要稍微計算一下，最後在接上 fully connected 層時要符合其 input shape 。
# Create CNN Model
class CNN_Model(nn.Module):
def __init__(self):
super(CNN_Model, self).__init__()
# Convolution 1 , input_shape=(1,28,28), output_shape=(16,24,24)
self.cnn1 = nn.Conv2d(in_channels=1, out_channels=16, kernel_size=5, stride=1, padding=0)
# activation
self.relu1 = nn.ReLU()
# Max pool 1, output_shape=(16,12,12)
self.maxpool1 = nn.MaxPool2d(kernel_size=2)
# Convolution 2, output_shape=(32,8,8)
self.cnn2 = nn.Conv2d(in_channels=16, out_channels=32, kernel_size=5, stride=1, padding=0)
# activation
self.relu2 = nn.ReLU()
# Max pool 2, output_shape=(32,4,4)
self.maxpool2 = nn.MaxPool2d(kernel_size=2)
# Fully connected 1, input_shape=(32*4*4)
self.fc1 = nn.Linear(32 * 4 * 4, 10)
def forward(self, x):
# Convolution 1
out = self.cnn1(x)
out = self.relu1(out)
# Max pool 1
out = self.maxpool1(out)
# Convolution 2
out = self.cnn2(out)
out = self.relu2(out)
# Max pool 2
out = self.maxpool2(out)
out = out.view(out.size(0), -1)
# Linear function (readout)
out = self.fc1(out)
return out
model = CNN_Model().to(device) # Create the CNN Model
optimizer = torch.optim.Adam(model.parameters(), lr = LR) # 選擇你想用的 optimizer(Adam)
summary(model, (1, 28, 28)) # 利用 torchsummary 的 summary package 印出模型資訊，input size: (1 * 28 * 28)
loss_func = nn.CrossEntropyLoss() # 選擇想用的 loss function(CrossEntropy)
input_shape = (-1, 1, 28, 28)
----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
Conv2d-1 [-1, 16, 24, 24] 416
ReLU-2 [-1, 16, 24, 24] 0
MaxPool2d-3 [-1, 16, 12, 12] 0
Conv2d-4 [-1, 32, 8, 8] 12,832
ReLU-5 [-1, 32, 8, 8] 0
MaxPool2d-6 [-1, 32, 4, 4] 0
Linear-7 [-1, 10] 5,130
================================================================
Total params: 18,378
Trainable params: 18,378
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.00
Forward/backward pass size (MB): 0.19
Params size (MB): 0.07
Estimated Total Size (MB): 0.27
----------------------------------------------------------------
7. 訓練模型
# 訓練 function
def fit_model(model, loss_func, optimizer, input_shape, num_epochs, train_loader, test_loader):
# Traning the Model
# 儲存訓練資訊的 List
training_loss, training_accuracy = [], []
validation_loss, validation_accuracy = [], []
for epoch in range(num_epochs):
# ---------------------------
# Training Stage
# ---------------------------
correct_train, total_train = 0, 0
for i, (images, labels) in enumerate(train_loader):
train, labels = images.view(input_shape).to(device), labels.to(device) # 取出 training data 以及 labels(轉 device 的型態)
optimizer.zero_grad() # 清空梯度
outputs = model(train) # 將訓練資料輸入至模型進行訓練 (Forward propagation)
train_loss = loss_func(outputs, labels) # 計算 loss
train_loss.backward() # 將 loss 反向傳播
optimizer.step() # 更新權重
# 計算訓練資料的準確度 (correct_train / total_train)
predicted = torch.max(outputs.data, 1)[1] # 取出預測的 maximum
total_train += len(labels) # 全部的 label 數 (Total number of labels)
correct_train += (predicted == labels).float().sum() # 全部猜中的個數 (Total correct predictions)
# 將 accuracy 和 loss 存入 list
train_accuracy = 100 * correct_train / float(total_train) # training accuracy (To cpu())
training_accuracy.append(train_accuracy.cpu())
training_loss.append(train_loss.data.cpu()) # training loss (To cpu())
# --------------------------
# Testing Stage
# --------------------------
correct_test, total_test = 0, 0
for images, labels in test_loader:
test, labels = images.view(input_shape).to(device), labels.to(device) # 取出 testing data 以及 labels(轉 device 的型態)
outputs = model(test) # 將測試資料輸入至模型進行測試 (Forward propagation)
val_loss = loss_func(outputs, labels) # 計算 loss
# 計算測試資料的準確度 (correct_test / total_test)
predicted = torch.max(outputs.data, 1)[1] # 取出預測的 maximum
total_test += len(labels) # 全部的 label 數 (Total number of labels)
correct_test += (predicted == labels).float().sum() # 全部猜中的個數 (Total correct predictions)
# 將 accuracy 和 loss 存入 list
val_accuracy = 100 * correct_test / float(total_test) # testing accuracy (To cpu())
validation_accuracy.append(val_accuracy.cpu())
validation_loss.append(val_loss.data.cpu()) # testing loss (To cpu())
# 顯現當前 Epoch 訓練情況
print('Train Epoch: {}/{} Traing_Loss: {} Traing_acc: {:.6f}% Val_Loss: {} Val_accuracy: {:.6f}%'.format(epoch+1, num_epochs, train_loss.data, train_accuracy, val_loss.data, val_accuracy))
return training_loss, training_accuracy, validation_loss, validation_accuracy
# 訓練模型
training_loss, training_accuracy, validation_loss, validation_accuracy = fit_model(model, loss_func, optimizer, input_shape, num_epochs, train_loader, test_loader)
8. 將每一個 epoch 的 Loss 以及 Training / Testing accuracy 紀錄下來並繪製成圖
# Loss
plt.plot(range(num_epochs), training_loss, 'b-', label='Training_loss')
plt.plot(range(num_epochs), validation_loss, 'g-', label='validation_loss')
plt.title('Training & Validation loss')
plt.xlabel('Number of epochs')
plt.ylabel('Loss')
plt.legend()
plt.savefig('loss.png')
plt.show()
# Accuracy
plt.plot(range(num_epochs), training_accuracy, 'b-', label='Training_accuracy')
plt.plot(range(num_epochs), validation_accuracy, 'g-', label='Validation_accuracy')
plt.title('Training & Validation accuracy')
plt.xlabel('Number of epochs')
plt.ylabel('Accuracy')
plt.legend()
plt.savefig('acc.png')
plt.show()
