This classic guide focuses on applying neural network models to Natural Language Processing (NLP). The first half of the book deals with the basics of machine learning with a teacher on linguistic data and the application of vector rather than symbolic representations of words. The abstraction of the computation graph is also discussed, which makes it easy to define and train arbitrary neural networks and underlies modern software neural network libraries. In the second part, more specialized neural network architectures are introduced, including one-dimensional convolutional networks, recurrent neural networks, conditional generation models and models with an attention mechanism.