New hands-on book makes deep learning and AI accessible

It all started in 2022 when Giovanni Volpe and colleagues gave a PhD course on AI for microscopy at ETH (Eidgenössische Technische Hochschule) in Zurich. The course was a great success, and the group decided to develop the extensive study material they had written for the students into a book. The author group consisted of Giovanni Volpe, Benjamin Midtvedt, Jesús Pineda, Henrik Klein Moberg, Harshith Bachimanchi, Joana B. Pereira and Carlo Manzo.

“Working on Depp Learning Crash Course was plenty of fun, even though it required a lot of reworking of the text. We had a lot of help from the publisher, No Starch Press. They revised every single word and line of code in the book several times,” says Giovanni Volpe.
In the book's acknowledgment, the authors describe their writing process as follows:

“Writing a book on deep learning is a lot like training a neural network. It's an endless cycle of tweaking parameters, second-guessing decisions, and resisting the urge to rage-quit and take up gardening instead, like when we decided to switch all the code from TensorFlow to PyTorch after completing the first draft of the book, because who doesn’t love rewriting everything from scratch?”

Groundbreaking for research

Deep learning is a form of AI that by using neural networks in several layers can be trained to do, for example, image recognition and text analysis. The technology has proven completely groundbreaking in natural science research - not least in physics - where projects often involve analysis and qualification of large amounts of research data. 

“Deep Learning Crash Course” was not written for already seasoned experts. On the contrary, it is intended to open the gates for curious programmers who want to learn how to build their own AI models from scratch. No prior knowledge of neural networks is required - the book is a concrete guide that helps the reader step-by-step to create and train their own models through practical exercises. 

"I want to emphasize that this is a book that can help readers without knowledge of deep learning or AI to understand and work with the very latest technology. Many of the projects in the book are actually based on recently published scientific articles, several of them from our research group," says Giovanni Volpe.
 

Text: Carolina Svensson
Photo: Giovanni Volpe