Mathematics Behind Neural Networks by T Aadhya (.ePUB)
File Size: 98.2 MB
Mathematics Behind Neural Networks: 400 Illustrated Exercises from Algebra to Transformers by T Aadhya
Requirements: .ePUB reader, 98.2 MB
Overview: Mathematics Behind Neural Networks: 400 Illustrated Exercises from Algebra to Transformers is a hands on workbook for readers who want to understand the mathematics inside modern neural networks, one computation at a time. If you have ever seen a neural network diagram and wanted to know what the numbers are actually doing, this book gives you the answer through worked exercises, clear notation, and structured practice. From scalars and vectors to matrix multiplication, activation functions, backpropagation, optimization, convolutional networks, recurrent networks, embeddings, and transformers, each chapter breaks the subject into concrete calculations you can perform by hand. This book is designed to help readers move beyond abstract explanations and build real mathematical fluency. Every problem asks you to compute something specific: a dot product, a forward pass, a loss value, a gradient, an attention weight, or a parameter update. The goal is not to guess the concept, but to calculate it clearly and correctly. This workbook takes that arithmetic seriously. Every one of the 400 problems in these pages asks you to compute something by hand—a forward pass, a gradient, an attention weight, a loss value. The problems are not puzzles or tricks. They are the exact computations that happen inside PyTorch and TensorFlow on every training step, made visible and tractable.
Genre: Non-Fiction > Tech & Devices
Free Download links: