With the tremendous growth and widespread adoption of deep learning and natural language processing (NLP), we often find ourselves referencing multiple books, research papers, and websites for the latest information and techniques. There is a growing need for a single comprehensive resource that provides a guide to the latest deep learning techniques for NLP and speech applications. With this goal in mind, my colleagues (Uday Kamath and Jimmy Whitaker) and I set out to write a textbook that can appeal to NLP practitioners, students, and anyone with an interest in recent deep learning NLP approaches and techniques.
We are pleased to announce our book Deep Learning for NLP and Speech Recognition published by Springer. Our mission was to write a book that serves as:
- A comprehensive resource that builds up from elementary deep learning, text, and speech principles to advanced state-of-the-art neural architectures
- A ready reference for deep learning techniques applicable to common NLP and speech recognition applications
- A useful resource on successful architectures and algorithms with essential mathematical insights explained in detail
- An in-depth reference and comparison of the latest end-to-end neural speech processing approaches
- A panoramic resource on leading-edge transfer learning, domain adaptation, and deep reinforcement learning architectures for text and speech
- A field guide on practical aspects of using these techniques with tips and tricks essential for real-world applications
- A hands-on tutorial on using Python-based libraries such as Keras, TensorFlow, and PyTorch to apply these techniques in the context of real-world case studies
Organization
The book is organized into three parts:
- Part 1 introduces the foundations of machine learning, NLP, and speech. It serves as an introduction or review of the basics of machine learning, text and speech processing.
- Part 2 introduces deep learning, word embeddings, and speech recognition fundamentals. It builds on the concepts from Part 1 to give readers a solid understanding of the building blocks for advanced techniques. It begins with an introduction to deep learning and continues to distributed representations, convolutional neural networks (CNNs), and recurrent neural networks (RNNs). Lastly, it covers automatic speech recognition (ASR) concepts.
- Part 3 explores advanced techniques and the latest research on attention mechanisms, transfer & multitask learning, reinforcement learning, and end-to-end deep learning applications for text and speech. This part is aimed at advanced students and practitioners. It starts with attention and memory-augmented networks. Next, it presents transfer learning, self-taught learning, and multi-task learning concepts and then continues to some of the most exciting recent advances in transfer learning and domain adaptation. It concludes with a discussion of advanced concepts in end-to-end speech recognition and deep reinforcement learning for text and speech.
Throughout the book, we present key concepts and leverage different frameworks and libraries to explore modern research and practical applications. Readers will find case studies at the end of each chapter that apply the concepts introduced in the chapter. These case studies are also available in GitHub repositories.
Conclusion
With the growing interest in deep learning and NLP for text and speech, it is difficult to find a definitive reference that encompasses current knowledge. We sought to create a comprehensive resource that covers both theory and practice. We hope readers will find this book useful as both a textbook and field guide to deep learning for NLP and speech applications.
Amazon: link to book
Springer: link to book
