16.03.2021 Views

Advanced Deep Learning with Keras

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Deep Neural Networks

if version == 2:

model = resnet_v2(input_shape=input_shape, depth=depth)

else:

model = resnet_v1(input_shape=input_shape, depth=depth)

The accuracy of ResNet v2 is shown in following table:

# Layers n % Accuracy on CIFAR10

(Original paper)

ResNet56 9 NA 93.01

ResNet110 18 93.63 93.15

% Accuracy on CIFAR10

(This book)

Table 2.3.1: The ResNet v2 architectures validated on the CIFAR10 dataset

In the Keras applications package, ResNet50 has been implemented as well with

the corresponding checkpoint for reuse. This is an alternative implementation but

tied to the 50-layer ResNet v1.

Densely connected convolutional

networks (DenseNet)

Figure 2.4.1: A 4-layer Dense block in DenseNet.

The input to each layer is made of all the previous feature maps.

DenseNet attacks the problem of vanishing gradient using a different approach.

Instead of using shortcut connections, all the previous feature maps will become

the input of the next layer. The preceding figure, shows an example of a dense

interconnection in one Dense block.

[ 62 ]

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!