Deep Learning with Python: A PyTorch Approach

Name: Deep Learning with Python: A PyTorch Approach
ISBN: 9787111637172

吴茂贵 / Yu Mingmin / Yang Benfa / 李涛 / 张粤磊China Machine Press2019

4.6

Word count: 198 words
Chapters: 22
Series: This book was written jointly by several engineers with years of hands-on experience in artificial intelligence and big data. As a companion volume to Deep Learning with Python: A TensorFlow Approach, it takes PyTorch as its main thread, systematically explaining deep learning across multiple dimensions: tool usage, technical principles, algorithm design, and case implementation. The book has 16 chapters in three parts. Part I (Chapters 1–4) covers PyTorch fundamentals: it first explains essential NumPy knowledge such as array creation, arithmetic operations, array reshaping, and broadcasting, then covers PyTorch installation and configuration, Tensor operations, autograd automatic differentiation and computational graphs, followed by the nn neural network toolbox and the data processing toolbox. Part II (Chapters 5–8) covers deep learning foundations: machine learning basics, visual processing basics, natural language processing basics, and generative deep learning, explaining the principles of classic models, training and tuning, and common techniques. Part III (Chapters 9–16) covers deep learning practice: through cases such as face detection and recognition, transfer learning, neural machine translation, generative models, Caffe2 model migration, adversarial attacks, reinforcement learning, and deep reinforcement learning, it demonstrates the complete engineering workflow from data preparation and model building to training and evaluation. Throughout, the book emphasizes starting from examples, progressing from easy to difficult, simplifying the complex, and letting figures do the talking, so that readers understand deep learning while coding hands-on.
ISBN: 9787111637172

Report an error Find this book

Summary

This is a hands-on textbook that binds "learning the framework" and "learning the principles" together. The author team has roughly twenty years of combined experience in BI, data mining, machine learning, and deep learning engineering, and what they write is pragmatic: they first move you from your old NumPy habits to Tensors, then take apart and explain thoroughly the parts you actually use every day—autograd, the nn toolbox, Dataset/DataLoader—rather than floating concepts in mid-air. Part II moves to the model level, advancing along three main lines in parallel—vision, NLP, and generative models—so readers quickly build the judgment of "which kind of network to use for which problem." The value of Part III lies in the completeness of its cases—face recognition, transfer learning, machine translation, generative models, adversarial attacks, reinforcement learning—covering basically the hottest directions in industry at the time, with each case showing how data, models, and training pipelines work together. The book has many figures and restrained formulas; where things get abstract, it tends to speak through diagrams and code, making it especially friendly to self-learners. It is a book to be read with code at your fingertips, with complete source code and data, suitable as your main deep learning textbook for a period of time rather than a bedside skim.

Editor's review

这是一本把「学框架」和「学原理」绑在一起讲的实战型教材。作者团队在 BI、数据挖掘、机器学习与深度学习工程上累计有二十年上下的经验，写出来的东西很务实：先把你从 NumPy 的老习惯过渡到 Tensor，再把 autograd、nn 工具箱、Dataset/DataLoader 这些真正每天要用的部件拆开讲透，而不是悬在半空谈概念。第二部分开始进入模型层面，视觉、NLP、生成式三条主线并行推进，读者能很快建立「什么问题该用哪类网络」的判断力。第三部分的价值在于案例的完整性——人脸识别、迁移学习、机器翻译、生成式模型、对抗攻击、强化学习，基本覆盖了当时工业界最热的方向，每个案例都能看到数据、模型与训练流程的配合。全书配图多、公式克制，遇到抽象的地方倾向用示意图和代码说话，对自学者尤其友好。它是一本需要边读边敲代码的书，配套源码与数据齐全，适合当作一段时间内的深度学习主教材，而不是枕边速览读物。

Who it's for

A good fit for

已掌握 Python 基本语法、想系统入门深度学习的开发者
需要在项目中落地 PyTorch 的算法工程师
做计算机视觉或自然语言处理课题的高校学生
从 TensorFlow 转向 PyTorch、想快速熟悉生态的从业者

Not a good fit for

完全不会 Python、也不了解基本编程概念的零基础读者
只想看概念综述、不打算动手写代码的读者
需要追踪 PyTorch 最新版本 API 变化的读者，建议看第2版

Reading tips

先补齐 NumPy 与线性代数基础，第1章不要跳过，后面 Tensor 操作依赖它。
边读边在 Jupyter Notebook 中复现代码，先跑通示例再改参数做消融实验。
第三部分案例可按兴趣选读，建议至少完整走通一个案例的数据到评估全流程。
注意书中 PyTorch 版本较老，遇到 API 报错先查官方文档的最新写法。

About the publisher

China Machine Press China · Founded in 1950

In addition to science and technology and professional publishing, the books in our catalog from China Machine Press also reflect a sustained focus on psychology, computing, management, finance and economics, and social sciences. In psychology, the catalog includes works by C. G. Jung and introductions to analytical psychology, as well as applied topics such as stress and health, and addiction and healing. In computing, it features Structure and Interpretation of Computer Programs (JavaScript Edition). Management and finance titles include Administrative Behavior (4th Edition), The Price of Inequality (Collector's Edition), and This Time Is Different: Eight Centuries of Financial Folly (Commemorative Edition). In the philosophy and history of science, there is The Worldview: The Third Edition. Relevant authors include C. G. Jung, William McGuire, Sonu Shamdasani, and Gabor Maté. Public information is limited; the above summarizes the publisher's profile based solely on the books in our catalog.

Preface
Part I PyTorch Fundamentals
Chapter 1 NumPy Fundamentals
Chapter 2 PyTorch Fundamentals
Chapter 3 PyTorch Neural Network Toolbox
Chapter 4 PyTorch Data Processing Toolbox
Part II Deep Learning Foundations
Chapter 5 Machine Learning Fundamentals
Chapter 6 Visual Processing Fundamentals
Chapter 7 Natural Language Processing Fundamentals
Chapter 8 Generative Deep Learning
Part III Deep Learning Practice
Chapter 9 Face Detection and Recognition
Chapter 10 Transfer Learning in Practice
Chapter 11 Neural Machine Translation in Practice
Chapter 12 Generative Models in Practice
Chapter 13 Caffe2 Model Migration in Practice
Chapter 14 A New Direction in AI: Adversarial Attacks
Chapter 15 Reinforcement Learning
Chapter 16 Deep Reinforcement Learning
Appendix A Changes in PyTorch 0.4
Appendix B The Latest Applications of AI Across Industries

FAQ

这本书和《动手学深度学习》有什么区别？

本书侧重 PyTorch 工具箱与工程案例的完整实现，覆盖人脸识别、机器翻译、对抗攻击、强化学习等方向；《动手学深度学习》更强调从零实现模型与原理推导。两者可以互补，前者偏工程进阶，后者偏入门与理论。

零基础可以直接读这本书吗？

不建议。书中假定读者会 Python，并且对数组运算和机器学习基本概念有一定了解。零基础读者建议先学 Python 与 NumPy，再补一点机器学习入门知识。

第1版和第2版该怎么选？

第2版于2023年由机械工业出版社出版，ISBN 为 9787117188802，全书19章，按新版 PyTorch 全面升级并新增注意力机制、目标检测、语义分割等内容。若使用较新的 PyTorch 版本，优先选第2版。

书中的代码能直接运行吗？

代码基于成书时的 Python 与 PyTorch 版本编写，直接运行大版本差异处可能报错，通常需要按当前版本的 API 做少量调整，并准备好 GPU 环境与数据集。

Comments

Find this book

This site is an index: you start “Search all sources” in your browser, and the entries it lists all point to third-party sites; please judge their availability yourself.

About the author

吴茂贵 Author

Wu Maogui — public information about this author is limited, and no authoritative account of their identity or background is available. Their writing direction can nonetheless be inferred from the two works held in the collection, both in computing and artificial intelligence: AIGC Principles and Practice: Learning Large Language Models, Diffusion Models, and Multimodal Models from Scratch, and From Python Basics to AI in Practice. The first introduces generative AI to beginners, explaining the principles of large language models, diffusion models, and multimodal models while guiding readers through hands-on practice; the second starts from Python and gradually moves toward applied artificial intelligence. Both emphasize a path from fundamentals to practical work, using accessible language suited to readers without prior background. Wu Maogui's writing thus tends to break complex technologies into actionable introductory material, combining conceptual explanation with exercises for newcomers to the field.

李涛 Author

Li Tao is an author whose works span a wide range of fields, with 27 titles held in the collection. The subjects he writes on are notably diverse: medical professional titles such as Clinical Psychopathology and the National County-Level Hospital Practical Handbook series volume Psychiatrist's Handbook, as well as clinical technical topics like Minimally Invasive Surgery for Diabetic Retinopathy; education and social issues in The Modernization of Chinese Rural Education; and practical tutorials on software and computing, including Adobe After Effects CC: The Master's Path. He has also written on science popularization and legal thinking, such as Perfect Symmetry: The Accidental Discovery of Fullerenes and 50 Lessons on Cultivating Citizens' Legal Thinking: Think Like a Lawyer, and contributed to applied writing such as the Technical Guidelines for Evaluating Healthy Enterprise Development. Because public information is limited, his personal biography, professional background, and identity remain unclear; from the distribution of his works, one can only see that he combines specialized writing with popular outreach, reaching readers in medicine, education, software, popular science, and law.

张粤磊 Author

Zhang Yuelei is an author whose publicly available information is currently limited; the name appears as the author credit on the programming and design book Database Application in Practice. Judging from the subject matter, the book addresses the practical use of database technology and belongs to hands-on technical writing rather than purely theoretical discussion. Books of this kind usually organize content around specific tasks and scenarios, emphasizing the translation of database knowledge into runnable examples and operational steps, and are suitable for readers who prefer learning by doing. Because reliable information about the author's background, professional training, and other writings is lacking, it is not possible to assess the full scope of his work or its specific influence among readers. Based on the available clues, Zhang Yuelei's writing focuses on database applications and programming practice and may serve readers interested in introductory and practical database exercises.