Methods and Techniques in Deep Learning: Advancements in mmWave Radar Solutions
Avik Santra, Souvik Hazra, Lorenzo Servadei, Thomas Stadelmayer, Michael Stephan and Anand Dubey
I remember freshman year physics reading Halliday, Resnick and Walker and laughing at the intro to the infamous problem No. 85:
“You are kidnapped by political science majors (who are upset because you told them political science is not a real science)…”
It’s approachable and engaging, and while the book is serious, it’s not too serious. It’s the Star Wars of textbooks.
Conversely, I just finished reading Methods and Techniques in Deep Learning. If Dr. Halladay and friends are George Lucas, then Dr. Santra and the crew that wrote this are Neal Stephenson. This book is dense, it’s heavy and you would argue with its veracity at your peril. The last book I would say that about had a red cover with the name Harrington embossed on it. So, this crew is in good company, but that also means their book is not for the faint of heart.
Read it through and it will give you impressive detail about radar and the various types of deep learning — a technology some will term AI, although I agree with the authors’ decision not to explicitly call it such — that are applicable. As frequencies go up and radar gets more precise, it’s easy to surpass human ability to parse large, fast, multi-dimensional datasets. Deep learning, on the other hand, is a fantastic way to see through noise, recognize patterns and boil that data to what we’re concerned with.
With all the detail, this is still a concise book; I found myself re-reading passages to make sure I unpacked what I wanted to. The authors have clearly spent considerable time wrestling various problems in the field and have documented it so you won’t have to repeat those efforts. With that said, it is likely useful for many things beyond what you will personally use it for, so I would recommend studying the table of contents and recognizing what your interests will be as opposed to reading it linearly.
Also, this is more of a reference than a primer, someone brand new to radar or deep learning will probably prefer something lighter to read first. It could be used as a supplement for a graduate school class involving mostly laboratory experimentation, but I generally wouldn’t outline a classroom course with it. Rather, I think the ideal case is to be in the library of researchers and engineers working on these types of problems, ready to catch them if they stumble. If you work in radar, deep learning or even machine learning, you owe it to yourself to have a copy of this book. The next time you stumble on a technical problem, or need a citation in a paper, you’ll be glad it’s there.
Reviewed by: Brian Rautio
ISBN 13: 978-1-119-91065-7
Pages: 336
To order this book, contact:
Wiley-IEEE Press (December 2022)
www.wiley.com