Natural Language Processing with TensorFlow

Key Features

• Focus on natural language processing with TensorFlow, thereby avoiding the traditional focus on computer vision
• Treats NLP as a field in its own right, and learn to process and evaluate large unstructured data sets consisting of text
• Learn to apply the TensorFlow toolbox to the most interesting field in artificial intelligence

Book Description

TensorFlow is the most important deep learning framework currently in existence. Deep Learning algorithms are the most important frontier in artificial intelligence, with natural language processing (NLP) providing much of the engineering required to understand and process the vast majority of data available to deep learning applications today. Natural Language Processing with TensorFlow teaches aspiring deep learning developers to cope with unstructured data, that is, text and audio, which make up a large part of currently available data streams.

Thushan Ganegedara starts out by explaining the inner workings of TensorFlow itself, and the proceeds with a family of algorithms allowing sequences of words to be turned into vectors, making them accessible to deep learning algorithms in the process. He then shifts gears somewhat by showing how classical deep learning algorithms like convolutional neural networks (CNN) and recurrent neural (RNN) networks can be applied to NLP. Long short-term memory are a variety of RNNs, used in the next chapter for text generation.

Thushan concludes the book with an overview and implementation of neural machine translation, a method relying on deep learning algorithms to achieve impressive results in machine translation.

What you will learn

• How to master NLP based on existing TensorFlow algorithms
• Build NLP applications
• Write automatic translation programs using neural machine translation algorithms
• Use classical deep learning algorithms to classify sentences
• Apply Long Short-Term Memory to text generation