Fundamentals of Deep Learning Models

Many deep learning textbooks either develop mathematical theory in isolation from the models it supports, or focus on code-level implementation without adequately addressing the underlying mathematics. Fundamentals of Deep Learning Models bridges this gap by interpreting the mathematical derivations of deep learning in direct connection with the architectures they underpin.Starting from linear models and progressing through MLPs, CNNs, and RNNs, the book provides detailed mathematical derivations for both forward and backward propagation in each architecture, together with the structural principles necessary for implementation. Building on these foundations, it then examines how these core components combine in advanced models—including Transformers, Neural Radiance Fields (NeRF), and Stable Diffusion—offering a unified mathematical and architectural analysis of each.Designed for graduate students and practitioners seeking a deeper understanding of modern deep learning, this book equips readers not only to implement these models but to comprehend and extend the mathematical reasoning at their core. Read more