Geometry of Deep Learning

The author is an Associate Professor at the Korea Advanced Institute of Science and Technology (KAIST). He has also been an IEEE Fellow since January 2020.

The goal of Geometry of Deep Learning is to give students insights into geometry that will help them grasp deep learning from a unified standpoint. Rather than defining deep learning as a technique for implementation, as is common in many existing deep learning publications, deep learning is described here as the ultimate form of signal processing procedures that may be conceived.

To back up this assertion, an overview of traditional kernel machine learning algorithms is offered, together with their benefits and drawbacks. Following a detailed discussion of the biological and computational building blocks of deep neural networks, the latest technologies such as attention, normalization, Transformer, BERT, GPT-3, and others are described. The emphasis here, too, is on the fact that beneath the intuition in these heuristic approaches is an important, beautiful geometric structure that allows for a systematic understanding. A unified geometric analysis is provided to comprehend the working mechanism of deep learning from high-dimensional geometry. Then, many types of generative models, such as GAN, VAE, normalizing flows, optimal transport, and so on, are explained from a unified geometric standpoint, demonstrating that they are derived from statistical distance-minimization concerns.

Geometry of Deep Learning can be used as an advanced deep learning textbook in universities or as a reference source for researchers interested in acquiring the latest deep learning algorithms and their underlying principles because it contains up-to-date information from both a practical and theoretical standpoint. Furthermore, the book was written for a codeshare course for students of both engineering and mathematics, so much of the information is interdisciplinary and will appeal to students from both disciplines.

Author: Jong Chul Ye

Link to buy: https://www.amazon.com/Geometry-Deep-Learning-Perspective-Mathematics/dp/981166045X/

Ratings: 5.0 out of 5 stars (from 3 reviews)

Best Sellers Rank: #118,911 in Books

#10 in Bioinformatics (Books)

#14 in Functional Analysis Mathematics

#17 in Differential Geometry (Books)