Book Introduction
This book is an introductory guide to learning deep learning by implementing core concepts from scratch. Rather than relying on libraries or frameworks, you can build a solid foundation in deep learning by writing code directly in Python. Complex concepts like backpropagation are explained visually through computational graphs for better understanding.
Author Introduction
Saito Goki is a graduate of the University of Tokyo and currently works in artificial intelligence research and development at a company. He is the author of the "Deep Learning from Scratch" series and has translated numerous programming books into Japanese.
Translator Introduction
Gae-ap-maeng-si (Lee Bok-yeon) is an IT specialist translator and editor. He has translated numerous IT technical books including the "Deep Learning from Scratch" series, and goes beyond simple translation by adding explanatory enhancements, code improvements, and various efforts to help readers understand. He conducts systematic and meticulous translation work based on rich experience in the IT technical book market.
Recommended for
- Beginners who want to learn deep learning step by step from the beginning
- Those who want to learn by implementing with actual working code
- Those who want to deeply understand from basic neural network principles to latest technologies (batch normalization, dropout, Adam, etc.)
This book is a series. At first, I thought 5 meant the 5th edition, but looking into it, each series has different topics.
Since they dig deep into big topics from scratch, you don't need to read each series in order. This 5th edition focuses on image generation models and Diffusion models, so if you're curious about other topics, please refer to the photo below.
It shows how to practice in Colab and various environments. 
In the case of other books that cover everything in one volume, for example, when explaining VAE, they explain it very briefly and move on, but in this fifth series, since the big topic focuses on Diffusion models, it explains more detailed implementation to applications.
In my case, I didn't study deep learning/machine learning sequentially, but was researching Diffusion model-related projects at work, so I already knew to some extent the content from VAE to Stable Diffusion covered in chapters 7-10 of this book. Reading this book was a great opportunity to learn more deeply about models, algorithms, and neural networks.
Even if you encounter it without prior knowledge, the book flow is excellent for learning, starting from Chapter 1's normal distribution to advanced image generation using Diffusion models.
Appendix
I discovered this by chance, so let me briefly introduce it - this is a novel created by the translator of this book, Gae-ap-maeng-si, based on 'Deep Learning from Scratch'. It's a novel that expresses deep learning topics and things that exist in the real world with added imagination.
Deep Learning Adventure Under the Sea
Conclusion
Being able to learn by following along line by line makes the book title "Deep Learning from Scratch" really resonate.
These days, many convenient deep learning tools and libraries are emerging, but I think this is a good book for those who want to learn the fundamental operating principles and basics step by step.
Although it's a somewhat difficult topic, it's still excellent that you can implement it through coding and the visual details are wonderful. Also, rather than explaining all necessary basic knowledge from the beginning at once, it appropriately explains what's needed in each of the 10 chapters when necessary.
"This review was written after receiving the book for Hanbit Media's <I am a Reviewer> activity."