VGG这个名字来源于论文作者所在的实验室Visual Geometry Group。其提出了可以通过重复使用简单的基础块来构建深度模型的思路。
VGG块(VGG-block)的结构:
1.单层或者多层的卷积层
2.单层池化层
import d2lzh as d2lfrom mxnet import gluon, init, ndfrom mxnet.gluon import nndef vgg_block(num_convs, num_channels): blk = nn.Sequential() for _ in range(num_convs): blk.add(nn.Conv2D(num_channels, kernel_size=3, padding=1, activation='relu')) blk.add(nn.MaxPool2D(pool_size=2, strides=2)) return blkVGG网络:
与AlexNet与LeNet一样,VGG网络由卷积层模块后接全连接层模块构成。卷积层模块串联数个vgg_block构成,其超参数由变量conv_arch定义,该变量指定了每个VGG块里卷积层个数和输出通道数。全连接模块则是由丢弃法的模块。
# 包含了5个vgg-block,总共8个卷积层,通道数慢慢增加到从64慢慢增加到512conv_arch = ((1, 64), (1, 128), (2, 256), (2, 512), (2, 512))# 使用了8个卷积层和3个全连接层,所以被称为VGG-11def vgg(conv_arch): net = nn.Sequential() # 卷积层部分 for (num_convs, num_channels) in conv_arch: net.add(vgg_block(num_convs, num_channels)) # 全连接层部分 net.add(nn.Dense(4096, activation='relu'), nn.Dropout(0.5), nn.Dense(4096, activation='relu'), nn.Dropout(0.5), nn.Dense(10)) return netnet = vgg(conv_arch)训练:
# 处于测试目的,将VGG网络的通道数缩小ratio = 4small_conv_arch = [(pair[0], pair[1] // ratio) for pair in conv_arch]net = vgg(small_conv_arch)# 训练lr, num_epochs, batch_size, ctx = 0.05, 5, 128, d2l.try_gpu()net.initialize(ctx=ctx, init=init.Xavier())trainer = gluon.Trainer(net.collect_params(), 'sgd', {'learning_rate': lr})train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=224)d2l.train_ch5(net, train_iter, test_iter, batch_size, trainer, ctx, num_epochs)参考文献:
动手学深度学习——阿斯顿张、深情的火车/p>