Deep Learning with Microsoft Cognitive Toolkit

Leverage the power of Deep learning with Microsoft's very own open source framework

About This Book

Build deployable solutions to tackle common deep learning problems.

Build high speed and efficient deep learning models using Microsoft Cognitive Toolkit.

Explore the various neural networks with the help of this comprehensive guide

Who This Book Is ForThis book is intended for data science professionals interested in deep learning and who would like to explore the new features of Microsoft CNTK. Basic Machine Learning and programming knowledge is assumed. Anyone looking for fetching deeper insights into their data with fast, open source tools will find this book quite helpful.

What You Will Learn

Learn basic concepts in deep learning

Know how to prepare data that can be consumed by CNTK for training

Discover how to train simple deep learning models with the Python API of CNTK

Evaluate trained CNTK models on new test data sets

Scale the execution of model training across multiple GPUs and multiple machines

Explore the common deep learning models for speech recognition, language understanding, image classification, etc.

Understand how to use CNTK to solve real-word problems such as face and emotion recognition, image segmentation, neural artistic style, image captioning, gaming, etc.

In DetailRight from setting up your neural network, this book will guide you to achive incredible computation speed with Microsoft Cognitive Tooklit. We will delve into machine learning aspects like speech comprehension, text to speech conversion, voice recognition, object recognition and many more. You will learn about the Network Description Language (NDL) and setting up the neural network. Further you will then explore the various deep learning architectures (CNNs, DNNs, RNNs etc) and how to use them to accelerate your computations. With practical examples, this book will teach you to work with time-series data and to create, train, and validate test sets. Later, you will understand how to work with larger datasets and dealing with heterogenous input data. You will also learn about the latest advances such as bindings with Python and C++.

By the end of this book, you will be able to migrate from other toolkits.