First contact with Deep Learning

Practical introduction with Keras

Table of content
Preface: the passion to teach
Preliminary considerations
Supercomputing, the heart of Deep Learning

1 A new disruptive technology is coming
1.1 Artificial Intelligence is changing our life
1.2 Artificial Intelligence, Machine Learning and Deep Learning
1.3 Why now?

2 Densely Connected Networks
2.1 Case of study: digit recognition
2.2 Perceptron
2.3 Softmax activation function

3 Case Study with Keras  
3.1 Data to feed a neural network
3.2 Densely connected networks in Keras
3.3 Basic steps to implement a neural network in Keras

4 Some basics about the learning process 
4.1 Learning process of a neural network
4.2 Activation functions
4.3 Backpropagation components
4.4 Model parameterization

5 Get started with Deep Learning hypeparameters
5.1 TensorFlow Playground
5.2 A binary classification problem
5.3 Hyperparameter setting basics

6 Convolutional Neural Networks 
6.1 Introduction to convolutional neural networks
6.2 Basic components of a convolutional neural network neuronal
6.3 Implementation of a basic model in Keras
6.4 Hyperparameters of the convolutional layer

Appendix: source code on GitHub

About the author
About Barcelona Supercomputing Center (BSC)
About Universitat Politècnica de Catalunya Barcelona Tech (UPC)
About Facultat d’Informàtica de Barcelona (FIB)


  • “” –> “”
  • pag 23: “Enrivonment set up” –> “Environment set up”
  • pag 51: “learning plase” –> “learning phase”

Course Slides (July 2018) 

  Paperback (16.64 €)

Compra este libro en Amazon.  Kindle Edition (8.00 €)

Support independent publishing: Buy this e-book on Lulu.  eBook PDF  (8.00 €)