[Book recommendation] 『If I Were Studying AI from Scratch Again』 by Kwon Gu-min
Translated by AIView original
A book by a coworker of mine comes out on October 26. It's 『내가 다시 AI를 처음 공부한다면』 by Kwon Gu-min (권구민), a researcher at work who is also an AI researcher.

There are plenty of intro books on AI, but most of them skip the concepts with "you don't need to know the math." This book goes the opposite way. It takes matrices, probability and derivatives and moves them one by one from equations into plain Korean sentences, starting with why the loss goes down. Further on, it dissects the Attention paper, derives the optimal discriminator for GANs by hand, and goes down to the entrance of flow matching. True to the title, it's a self-study roadmap for what order to go in if you were starting over from scratch.
I've watched the author up close. Whenever a new paper comes out, he understands the concepts and the math faster than anyone and then implements them in code right away. He's the kind of engineer who shows that getting the equations right makes implementation faster, so it makes sense that the book tackles the math head-on. The foreword was also written by Sunny Sanyal, the author of a paper he reproduced and a former Student Researcher at Google DeepMind.

This one is personally exciting for me too. When I was first learning VLA, I relied a lot on video material, and I wished there were a written resource that laid things out step by step in the same order. When I heard about this book, I thought, "oh, that's exactly the book."

I'd recommend it to anyone who wants to study AI properly, especially people who want to rebuild their foundations before moving on to Physical AI or generative models. It's up for pre-order now.
Comments 0
No comments yet
Be the first to comment.