Multimodal Deep Learning with Tensorflow: Translate mathematics into robust TensorFlow applications with Python

Become an expert in designing and deploying TensorFlow models that generate insightful predictions with the power of deep learning. About This Book * Understand the fundamentals of multimodal deep learning *Gain hands-on experience in implementing graph convolutional networks in TensorFlow *Get comfortable in applying a diverse set of deep architectures beyond traditional ensembles Who This Book Is For If you are a programmer who doesn't have a strong mathematical background and little practice in building advanced TensorFlow applications, this book will prove very beneficial for you. This book will also help Ph.D. students by providing them the necessary tools to build their own solutions. Moderate knowledge of TensorFlow is presumed because building such applications requires strong cognitive ability and basic programming skills. What You Will Learn * Detect emotions using facial expressions, voices, and gestures *Caption images using pictures, text, and voice as modalities *Build and test enhanced multiresolution image style transfer model *Use the multihop LINE algorithm to embed graph vertices in word2vec way *Implement a simply multiview GCN to predict traffic congestion *Refresh basic and intermediate concepts of graph theory In Detail Multimodal Deep learning (MDL) enjoys a wide spectrum of applications ranging from e-commerce and security screening to complicated healthcare applications. Through this book, you'll gain access to unique material in multimodal deep learning. Starting from simple image-based emotion recognition as a running example, you'll add more modalities, such as voice and gestures to illustrate how they improve the performance of the model. You'll also understand the mathematical background of emotion recognition and implement emotion recognition pipeline in TensorFlow. As the chapters progress, you'll learn methods to run machine learning algorithms on graphs. Through various applications, such as Parkinson disease identification and taxi ride demand prediction, you'll explore generalizing convolutional networks and how they can be used. By the end of the book, you'll have enough mathematical background and TensorFlow knowledge to implement your own multimodal deep learning applications.

About the Author :
Andrew Bout is focused on challenging machine learning problems for more than 5 years. He builds patented solutions in computer vision and NLP areas using TensorFlow, Keras, and other frameworks at Samsung and Kaspersky Lab. He was the strongest contributor to the first framework for model- and data-parallel deep networks training, built by Samsung. At leisure, he competes at Kaggle. Alexey Miasnikov is a machine learning hacker since 2004, building full-text search systems and hotel ranking in large companies such as Agoda or Samsung. He also has entrepreneur spirit and recently set up his own consulting startup, helping companies solve complex data science problems involving big data, guiding full cycle from a green field to production at scale. An avid traveler, specializing in uncovering unspoiled parts of a planet with his wife and three children. Gianluca Ortolani is strongly committed to machine learning with major focus on deep reinforcement learning, he started exploring NLP solutions while doing research on knowledge base management. He moved to real time CNN and image processing driven by his passion for computer vision. During the last years, he developed his expertise in TensorFlow and Spark, processing data at scale as a machine learning manager in Agoda.