Welcome to Kolmogorov Arnold Network (KAN) documentation!

This documentation is for the paper “KAN: Kolmogorov-Arnold Networks” and the github repo. Kolmogorov-Arnold Networks, inspired by the Kolmogorov-Arnold representation theorem, are promising alternatives of Multi-Layer Preceptrons (MLPs). KANs have activation functions on edges, whereas MLPs have activation functions on nodes. This simple change makes KAN better than MLPs in terms of both accuracy and interpretability.

Installation

Installation via github

git clone https://github.com/KindXiaoming/pykan.git
cd pykan
pip install -e .
# pip install -r requirements.txt # install requirements

Installation via PyPI

pip install pykan

Requirements

Get started

Quickstart: Hello, KAN!
KANs in Action: API Demos, Examples
API (advanced): API.

Contents: