This tutorial will cover the Keras deep learning library, by François Chollet, which is becoming the lingua franca to build Deep Learning architectures, and work on the built model's training and evaluation, on top of many of the high performance counterparts. During this tutorial, we will bring some of the most cutting edge applications of Machine Learning, and will explain how multilayer models can be built layer by layer, and how the trained models can be queried and visualized.

Themes Description:

1. Multi-layer Feed Forward Networks: We will explore the basic elements of the Keras api, especially the Sequential Object.

2. Layer exploration: We will use the Keras api to explore a determined model.

3. Convolutional Networks :This notebook will teach how to build CNN (Convolutional Neural Networks), using Convolutional, Pooling and DropOut layers

4. Transfer Learning: Will discover the Keras implementation of some foundational Deep Convolutional architecures, and use transfer learning to apply them to new problems.

5. Neural network visualization with Quiver: We will explore the layer of complex neural networks, and graphically analyze the different stages in the succesive stacked layers.

6. Recursive Neural Networks: We will use keras.layers.convolutional_recurrent to predict values of a time series.

7. Implementation details: Will discuss how to expose a determined model through an REST API.

Software requirements:

• Python 3.x.
• numpy version 1.10 or later: http://www.numpy.org/
• scipy version 0.16 or later: http://www.scipy.org/
• pandas version 0.16 or later: http://pandas.pydata.org
• scikit-learn version 0.15 or later: http://scikit-learn.org
• matplotlib version 1.5 or later: http://matplotlib.org/
• keras version 2.0 or later: http://keras.io
• tensorflow version 1.2 or later: https://www.tensorflow.org
• __jupyter__version 4 or later, with notebook support
• pyyaml
• hdf5 and h5py