| Chen Daxin
AI Technology Review today introduces Zhu Jingbo's new book: Machine Translation: Statistical Modeling and Deep Learning Methods (2nd Edition)-the basic chapter.
This book comprehensively reviews the technical development of machine translation in recent 30 years, and introduces the technical methods of machine translation around the theme of machine translation modeling.
The author introduces:
draw
Note: Zhu Jingbo (left) and Xiao Tong (right)
Zhu Jingbo, Professor, Mavericks Translator & Founder; Chairman, Professor of Artificial Intelligence Department, School of Computer Science, Northeastern University, doctoral supervisor.
Xiao Tong, associate professor, CEO of calf translation, deputy director of Natural Language Processing Laboratory of Northeastern University.
In their writing, the two academic giants strive to explain the basic model of machine translation with simple language and concise examples, and explore the relevant technical frontiers. It will also involve a lot of practical experience, including many details of machine translation system development.
From this point of view, this book is not only a theoretical book, but also combines the application of machine translation, providing readers with many specific ideas for the landing of machine translation technology.
This book has a PDF version, which can be downloaded and read online for free: mtbook/index.html
It is worth mentioning that the code of this book is also open source: NiuTrans/MTBook.
This book is written in Tex, with excellent typesetting, exquisite illustrations, excellent reading experience and people can't put it down.
The following two pages of illustrations are randomly selected:
More illustrations:
This book is divided into four parts, and each part consists of several chapters. The order of each chapter refers to the time sequence of the development of machine translation technology, and considers the internal logic of machine translation knowledge system.
The main relationships of each part are as follows:
The first part is the basic knowledge of this book, including statistical modeling, language analysis, machine translation evaluation and so on.
After the first chapter introduces the history and present situation of machine translation, the second chapter expounds the idea of statistical modeling through the task of language modeling, which will also serve as the basis of subsequent machine translation models and methods.
The third chapter focuses on the lexical and syntactic analysis methods involved in machine translation, aiming at paving the way for the subsequent use of related concepts and further demonstrating the application of statistical modeling ideas in related issues. The fourth chapter is relatively independent, and systematically introduces the evaluation method of machine translation results, which is also the preparatory knowledge needed for machine translation modeling and system design.
The second part mainly introduces the basic model of statistical machine translation. The fifth chapter is the basis of the whole machine translation modeling.
Chapter six further introduces the concepts of distortion and output rate, and gives the related translation model, which will be involved in later chapters. Chapters 7 and 8 introduce the model based on phrase and syntax respectively. They are all classic models of statistical machine translation, and their thoughts also constitute the most essential part of the growth of machine translation.
The third part mainly introduces the neural machine translation model, which is also a hot topic in machine translation research in recent years.
The ninth chapter introduces the basic knowledge of neural network and deep learning to ensure the integrity of the knowledge system of this book. At the same time, the ninth chapter also introduces the language model based on neural network, and its modeling idea is widely used in neural machine translation.
In chapter 10, 1 1 and 12, three classic models of neural machine translation are introduced respectively, from the initial model based on ring network to the latest model according to the time sequence proposed by the models. This will also affect the encoder? Classical methods and technologies such as decoder framework and attention mechanism are introduced.
The fourth part will further discuss the cutting-edge technology of machine translation, mainly neural machine translation. This part is currently being written and will meet with readers soon.
Book catalogue:
More directories:
This book can be used for senior undergraduates and graduate students majoring in computer science, and can also be used as a reference for researchers in the field of natural language processing, especially in the field of machine translation. In addition, the theme of each chapter of this book is very clear and the content is relatively concentrated. Therefore, readers can also use each chapter as learning materials on a specific topic.
This book is a course aimed at systematically introducing statistical modeling and deep learning methods of machine translation. Its contents have been compiled into a book, which can be used by senior undergraduates and graduate students majoring in computer science, and can also be used as reference materials for researchers related to natural language processing, especially machine translation.
What are you waiting for? Look at it first!
If you have any questions about the self-taught/adult-taught examination, don't know the contents of the test sites for self-taught/adult-taught examination, and don't know the local policies for self-taught/adult-taught examination, click on Mr. official website at the bottom to get the review materials for free: /xl/