Practical Spiking Neural Networks

A hands-on introduction to Spiking Neural Networks (SNNs) that teaches you to design, build, train, and deploy neuromorphic systems - both in simulation and on hardware.

What is this book about?¶

SNNs are biologically inspired neural networks that use discrete spikes for computation, making them inherently sparser and more energy-efficient than conventional ANNs when deployed on neuromorphic hardware. Computationally, SNNs are also more expressive because they combine numerical calculus (in the continuous domain $\mathbb{R}$) with discrete logic (in the discrete domain $\mathbb{N}$) Neumann, 2012.

This first-of-its-kind, community-driven and open-source book on SNNs bridges theory and practice with interactive examples, guiding you from SNN fundamentals through training and optimization methods to neuromorphic hardware deployments. We cover three topics:

1. Fundamentals: Covered in Topic 1: Foundations of SNNs

2. Training and Optimization: Covered in Topic 2: Training SNNs

3. Neuromorphic Hardware: Covered in Topic 3: Deploying SNNs

How do I get the book?¶

The book comes in two formats and both are free:

• Online at snnbook.net: the recommended way to read it. Only the website has the interactive code examples and visualizations, and it always shows the latest content.

• As a PDF: every release ships a snnbook.pdf on the releases page. The latest PDF is a static snapshot, so the interactive parts are rendered as still images.

What is in version 0.8?¶

All three topics have opened, each with a first set of draft chapters:

• Topic 1 - Foundations of SNNs: what a spiking neuron is (point neuron models, spiking vs. artificial neurons), encoding and decoding (rate encoding, temporal encoding, decoding), synaptic plasticity, and how to build networks out of spiking neurons.

• Topic 2 - Training SNNs: the importance of training, how to assignment credit in SNNs, and surrogate gradient training.

• Topic 3 - Deploying SNNs: why the hardware target matters, and hardware design principles.

• Appendix: discretization of point neuron models.

Chapters on exact gradients, meta learning, biologically inspired and evolutionary training, ANN-to-SNN conversion, training optimization, the neuromorphic compilation toolchain, interoperability, and performance modelling are in preparation and will appear in later releases.

Who is writing this?¶

The book is actively written by a group of more than 10 researchers who range from late-stage PhD candidates to scientists with decades of neuromorphic experience. The authors provide this as an open-source resource whose contents are reviewable by all, and thus vetted by the wider neuromorphic community for correctness.

The book is edited by:

• Ramashish Gaurav

• Jens Egholm Pedersen

• Petrut A. Bogdan

Contributing¶

We welcome improvements and pull requests. If you are interested to contribute by writing or reviewing, read more in the contribution guidelines and reach out to us on the Open Neuromorphic Discord server.

Citing¶

This book can be cited either as a whole or as individual chapters. We very much encourage citing specific chapters to credit the contributors who helped your research. The chapter-wise citations can be found in the web version.

@book{snnbook2026,
  editor    = {Gaurav, Ramashish and Pedersen, Jens Egholm and Bogdan, Petrut},
  title     = {{Practical Spiking Neural Networks}},
  publisher = {Open Neuromorphic},
  year      = {2026},
  edition   = {Version 0.8},
  url       = {https://snnbook.net},
}

License¶

The book is open source under CC BY 4.0.