Book Image

Advanced Machine Learning with Python

Book Image

Advanced Machine Learning with Python

Overview of this book

Designed to take you on a guided tour of the most relevant and powerful machine learning techniques in use today by top data scientists, this book is just what you need to push your Python algorithms to maximum potential. Clear examples and detailed code samples demonstrate deep learning techniques, semi-supervised learning, and more - all whilst working with real-world applications that include image, music, text, and financial data. The machine learning techniques covered in this book are at the forefront of commercial practice. They are applicable now for the first time in contexts such as image recognition, NLP and web search, computational creativity, and commercial/financial data modeling. Deep Learning algorithms and ensembles of models are in use by data scientists at top tech and digital companies, but the skills needed to apply them successfully, while in high demand, are still scarce. This book is designed to take the reader on a guided tour of the most relevant and powerful machine learning techniques. Clear descriptions of how techniques work and detailed code examples demonstrate deep learning techniques, semi-supervised learning and more, in real world applications. We will also learn about NumPy and Theano. By this end of this book, you will learn a set of advanced Machine Learning techniques and acquire a broad set of powerful skills in the area of feature selection & feature engineering.
Table of Contents (17 chapters)
Advanced Machine Learning with Python
Credits
About the Author
About the Reviewers
www.PacktPub.com
Preface
Chapter Code Requirements
Index

Stacked Denoising Autoencoders


While autoencoders are valuable tools in themselves, significant accuracy can be obtained by stacking autoencoders to form a deep network. This is achieved by feeding the representation created by the encoder on one layer into the next layer's encoder as the input to that layer.

Stacked denoising autoencoders (SdAs) are currently in use in many leading data science teams for sophisticated natural language analyses as well as a hugely broad range of signals, image, and text analysis.

The implementation of a SdA will be very familiar after the previous chapter's discussion of deep belief networks. The SdA is used in much the same way as the RBMs in our deep belief networks were used. Each layer of the deep architecture will have a dA and sigmoid component, with the autoencoder component being used to pretrain the sigmoid network. The performance measure used by a stacked denoising autoencoder is the training set error, with an intensive period of layer-to-layer...