Press "Enter" to skip to content

我复现的第一个神经网络: LeNet

本站内容均来自兴趣收集,如不慎侵害的您的相关权益,请留言告知,我们将尽快删除.谢谢.

 

学习深度学习已经有小一年的时间,看了很多视频和书本内容,学习了很多代码,可始终感觉认知不够扎实。结合李沐老师的视频课程,我决定在本博客中介绍下复现LeNet的过程。代码基于pycharm2021平台,选用python3.8版本+pytorch1.12.1+cu116。基本上把各个包的版本都刷到最新版本,以方便后续的网络升级和向后兼容。

 

1. LeNet简介

 

LeNet网络 [1] 由时任AT&T贝尔实验室的研究员Yann LeCun提出,可以被看作是卷积神经网络的开山之作。之所以选用LeNet作为尝试复现的第一个神经网络,是因为该网络本身的结构简单清晰,便于理解。作为早期成功应用于银行和邮政系统的实用型卷积神经网络,LeNet的结构足够经典,其中很多思想传承至今。因此,LeNet作为深度网络代码复现的一个经典案例,十分恰当。

 

我们首先回顾下LeNet的基本结构。输入是一个32*32的单通道图片 (更新版本的minist数据集的图片尺寸可能减到28*28,那幺在卷积的时候需要padding以保证卷积后的特征图为28*28),之后使用一个卷积层,变换出6通道的28*28的C1 feature map;加一步pooling,由28*28的feature压到14*14。之后按照相同的步骤,压出一个16通道的5*5的feature map,最后加两个全连接层,并输出10个元素组成的向量,以判断输入数字的类别。可以看到,整个结构是非常清晰,便于理解的。

 

2. LeNet实现

 

LeNet的网络搭建如下:

 

import torch
from torch import nn
from d2l import torch as s2l
class Reshape(torch.nn.Module):
    def forward(self, x):
        return x.view(-1,1,28,28)
net = torch.nn.Sequential(Reshape(),
                          nn.Conv2d(1,6,kernel_size = 5,padding=2),
                          nn.Sigmoid(),
                          nn.AvgPool2d(kernel_size=2,stride=2),
                          nn.Conv2d(6,16,kernel_size=5),
                          nn.Sigmoid(),
                          nn.AvgPool2d(kernel_size=2,stride=2),
                          nn.Flatten(),
                          nn.Linear(16*5*5, 120),
                          nn.Sigmoid(),
                          nn.Linear(120, 84),
                          nn.Sigmoid(),
                          nn.Linear(84, 10))
X = torch.rand(size = (1,1,28,28),dtype=torch.float32)
for layer in net:
    X = layer(X)
    print(layer.__class__.__name__,'output shape:\t',X.shape)

 

可以看到,整个网络的实现还是比较简单的。这里,按照李沐老师视频的介绍,我们给一个随机的输入,来输出网络中各个层对于输入数据的改变,结果如下:

 

Reshape output shape:        torch.Size([1, 1, 28, 28])

 

Conv2d output shape:          torch.Size([1, 6, 28, 28])

 

Sigmoid output shape:         torch.Size([1, 6, 28, 28])

 

AvgPool2d output shape:     torch.Size([1, 6, 14, 14])

 

Conv2d output shape:          torch.Size([1, 16, 10, 10])

 

Sigmoid output shape:         torch.Size([1, 16, 10, 10])

 

AvgPool2d output shape:     torch.Size([1, 16, 5, 5])

 

Flatten output shape:           torch.Size([1, 400])

 

Linear output shape:            torch.Size([1, 120])

 

Sigmoid output shape:         torch.Size([1, 120])

 

Linear output shape:            torch.Size([1, 84])

 

Sigmoid output shape:         torch.Size([1, 84])

 

Linear output shape:            torch.Size([1, 10])

 

在确定网络结构后,我们提取测试数据。这里,我们使用Fashion-MNIST数据集来训练和测试网络的性能。数据提取代码如下:

 

batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size = batch_size)

 

使用GPU计算模型在数据集上的精度:

 

def evaluate_accuracy_gpu(net, data_iter,device=None):
    if isinstance(net, torch.nn.Module):
        net.eval()
        if not device:
            device = next(iter(net.parameters())).device
    metric = d2l.Accumulator(2)
    for X,y in data_iter:
        if isinstance(X,list):
            X = [x.to(device) for x in X]
        else:
            X = X.to(device)
        y = y.to(device)
        metric.add(d2l.accuracy(net(X),y), y.numel())
    return metric[0]/metric[1]

 

添加训练函数的完整代码:

 

import torch
from torch import nn
from d2l import torch as d2l
class Reshape(torch.nn.Module):
    def forward(self, x):
        return x.view(-1,1,28,28)
def evaluate_accuracy_gpu(net, data_iter,device=None):
    if isinstance(net, torch.nn.Module):
        net.eval()
        if not device:
            device = next(iter(net.parameters())).device
    metric = d2l.Accumulator(2)
    for X,y in data_iter:
        if isinstance(X,list):
            X = [x.to(device) for x in X]
        else:
            X = X.to(device)
        y = y.to(device)
        metric.add(d2l.accuracy(net(X),y), y.numel())
    return metric[0]/metric[1]
def train_ch6(net, train_iter, test_iter, num_epochs, lr ,device):#lr: learning rate
    """train a model woth GPU"""
    def init_weights(m):
        if type(m) == nn.Linear or type(m) == nn.Conv2d:
            nn.init.xavier_uniform_(m.weight)
    net.apply(init_weights)
    print('training on', device)
    net.to(device)
    optimizer = torch.optim.SGD(net.parameters(),lr=lr)
    loss = nn.CrossEntropyLoss()
    animator = d2l.Animator(xlabel = 'epoch', xlim = [1, num_epochs],
                            legend = ['train loss', 'train acc', 'test acc'])
    timer, num_batches = d2l.Timer(),len(train_iter)
    for epoch in range(num_epochs):
        metric = d2l.Accumulator(3)
        net.train()
        for i, (X, y) in enumerate(train_iter):
            timer.start()
            optimizer.zero_grad()
            X, y = X.to(device), y.to(device)
            y_hat = net(X)
            l = loss(y_hat, y)
            l.backward()
            optimizer.step()
            metric.add(l * X.shape[0], d2l.accuracy(y_hat, y), X.shape[0])
            timer.stop()
            train_l = metric[0] / metric[2]
            train_acc = metric[1] / metric[2]
        if(i + 1) % (num_batches // 5) == 0 or i == num_batches - 1:
            animator.add(epoch + (i + 1)/ num_batches,(train_l, train_acc, None))
        test_acc = evaluate_accuracy_gpu(net, test_iter)
        animator.add(epoch + 1, (None, None, test_acc))
        print('Epoch:', epoch)
        print(f'loss {train_l:.3f}, train acc {train_acc:,.3f},' f'test acc {test_acc:.3f}')
        print(f'{metric[2] * num_epochs / timer.sum():.1f} examples/sec' f'on {str(device)}')
    print(f'loss {train_l:.3f}, train acc {train_acc:,.3f},' f'test acc {test_acc:.3f}')
    print(f'{metric[2] * num_epochs/timer.sum():.1f} examples/sec' f'on {str(device)}')
    print('finished')
def main():
    net = torch.nn.Sequential(Reshape(),
                              nn.Conv2d(1, 6, kernel_size=5, padding=2),
                              nn.Sigmoid(),
                              nn.AvgPool2d(kernel_size=2, stride=2),
                              nn.Conv2d(6, 16, kernel_size=5),
                              nn.Sigmoid(),
                              nn.AvgPool2d(kernel_size=2, stride=2),
                              nn.Flatten(),
                              nn.Linear(16 * 5 * 5, 120),
                              nn.Sigmoid(),
                              nn.Linear(120, 84),
                              nn.Sigmoid(),
                              nn.Linear(84, 10))
    batch_size = 256
    train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size=batch_size)
    lr, num_epochs = 0.9, 10
    train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())
if __name__ == '__main__':
    main()

 

3. 实验结果

 

打印结果:

 

training on cuda:0

 

Epoch: 0

 

loss 2.317, train acc 0.102,test acc 0.100

 

174566.9 examples/secon cuda:0

 

Epoch: 1

 

loss 1.383, train acc 0.459,test acc 0.580

 

139471.3 examples/secon cuda:0

 

Epoch: 2

 

loss 0.857, train acc 0.661,test acc 0.652

 

115809.0 examples/secon cuda:0

 

Epoch: 3

 

loss 0.718, train acc 0.716,test acc 0.701

 

99568.9 examples/secon cuda:0

 

Epoch: 4

 

loss 0.648, train acc 0.748,test acc 0.752

 

87336.1 examples/secon cuda:0

 

Epoch: 5

 

loss 0.590, train acc 0.770,test acc 0.776

 

77399.1 examples/secon cuda:0

 

Epoch: 6

 

loss 0.550, train acc 0.787,test acc 0.781

 

69605.1 examples/secon cuda:0

 

Epoch: 7

 

loss 0.515, train acc 0.800,test acc 0.793

 

63230.5 examples/secon cuda:0

 

Epoch: 8

 

loss 0.485, train acc 0.816,test acc 0.799

 

57836.1 examples/secon cuda:0

 

Epoch: 9

 

loss 0.459, train acc 0.829,test acc 0.761

 

53456.0 examples/secon cuda:0

 

loss 0.459, train acc 0.829,test acc 0.761

 

53456.0 examples/secon cuda:0

 

动态曲线图:

 

 

注:如果动画无法显示,参考博客: 无法显示动图怎幺办?

 

Reference

 

[1] LeCun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.

Be First to Comment

发表回复

您的电子邮箱地址不会被公开。