Welcome
License
1
Introduction
1.1
Introduction
I Using Torch
Introduction
2
A simple neural network in R
2.1
What’s in a network?
2.1.1
Gradient descent
2.1.2
From linear regression to a simple network
2.2
A simple network
2.2.1
Simulate data
2.2.2
Initialize weights and biases
2.2.3
Training loop
2.2.4
Complete code
3
Modifying the simple network to use torch tensors
3.1
Tensors
3.1.1
Creation
3.1.2
Conversion to built-in R data types
3.1.3
Indexing and slicing tensors
3.1.4
Reshaping tensors
3.1.5
Operations on tensors
3.2
Running on GPU
3.3
Broadcasting
3.4
Simple neural network using
torch
tensors
4
Using autograd
4.1
Automatic differentiation with
autograd
4.2
The simple network, now using
autograd
5
Using torch modules
5.1
Modules
5.1.1
Base modules (“layers”)
5.1.2
Container modules (“models”)
5.2
Simple network using modules
6
Using
torch
optimizers
6.1
Losses and loss functions
6.2
Optimizers
6.3
Simple network: final version
II Image Recognition
Introduction
7
Classifying images
7.1
Data loading and preprocessing
7.2
Model
7.3
Training
7.4
Test set accuracy
8
Brain Image Segmentation with U-Net
8.1
U-Net
8.2
Brain image segmentation
8.3
Data
8.4
Dataset
8.5
Model
8.6
Optimization
8.7
Training
8.8
Evaluation
III Time series
Introduction
9
Torch transformer modules
10
Sequence-to-sequence models with attention
11
Torch transformer modules
IV Natural language processing
Introduction
12
Sequence-to-sequence models with attention
13
Torch transformer modules
V Tabular data
Introduction
14
Handling categorical data
14.1
Agenda
14.2
Dataset
14.3
Model
14.4
Training
14.5
Evaluation
14.6
Making the task harder
14.7
A look at the hidden representations
VI Generative deep learning
Introduction
15
Generative adversarial networks
15.1
Dataset
15.2
Model
15.2.1
Generator
15.2.2
Discriminator
15.2.3
Optimizers and loss function
15.3
Training loop
15.4
Artifacts
16
Variational autoencoders
16.1
Dataset
16.2
Model
16.3
Training the VAE
16.4
Latent space
VII Deep learning on graphs
Introduction
VIII Probabilistic deep learning
Introduction
IX Private and secure deep learning
Introduction
X Research topics
Introduction
References
Applied deep learning with torch from R
10
Sequence-to-sequence models with attention