Deep learning. Ready-made solutions
Deep learning technology is not as complex as many believe. Until recently, it took years to study it, but with the advent of frameworks such as eras and TensorFlow, software engineers with no experience in this field can quickly start creating working applications. Thanks to the ready-made examples given in the book, you will learn how to solve problems related to the classification and generation of text, images and music. Each chapter describes several solutions that are combined into a single project, for example, an application that implements the training of a music recommendation system. There is also a chapter describing techniques that, if necessary, will help to debug the neural network. All examples are written in Python and are available as a set of notebooks. The main topics of the book are: - Using vector representations of words to calculate the similarity of texts - Building a recommendation system of films based on Wikipedia links - Visualization of the internal states of a neural network - Creating a model recommending emojis for text fragments - Reuse of pre-trained networks to create a reverse image search service - Generating pictograms using generative-adversarial networks (GAN), auto-encoders and recurrent networks (RNN) - Recognition of musical genres and indexing of song collections
No reviews found
