Format¶
This SNN book currently exists as a live website; it will also be made downloadable as a staic PDF. The live website will have interactive code examples that can be run on your browser. We recommend using the live website to interact with code examples and visualizations rather than just plainly reading the book.
Structure:¶
As mentioned before, this book is organized into three topics, each consisting of multiple planned chapters.
Foundations of SNNs: An introduction to biological and spiking neurons, encoding and decoding, concepts of plasticity, and methods to build a typical Spiking Neural Network (SNN) from scratch.
Training SNNs: An introduction to the concepts of training and optimizing SNN architectures, including surrogate gradient descent, meta learning, biologically inspired and evolutionary methods, and of course, the ANN-to-SNN conversion method, etc.
Deploying SNNs: An introduction to neuromorphic hardware design principles, neuromorphic-deployment compilation toolchain, platform-specific deployment examples and interoperability, and the neuromorphic sensors.
Each topic is self-contained with interactive examples. You can read them up sequentially or jump to specific topics as needed.
Prerequisites:¶
To read this book, you only need a basic understanding of calculus and linear algebra. If you’re new to spiking neurons or SNNs, start with Topic 1. If you’re familiar with the foundational concepts, use the later topics on training or deployment as a practical reference.
Learning approach:¶
Take time to read, and run and modify the interactive code examples in your browser; be curious! Understanding the semantics builds stronger intuition than memorizing solutions. The goal of this book is to give you practical and working knowledge of SNNs that you can apply immediately! Happy Reading!