TNLearn: Task-based Neurons for Learning

---

TNLearn

tnlearn , a Python package for implementing task-based neurons that aims to be easy to use, versatile for different data, and performant on different tasks.

• Easy-to-use: It provides a zero-barrier package for novices and a state-of-the-art benchmark for experienced researchers. Users can get results in 8 lines of code.

• Versatile: TNLearn constructs task-based neurons which are versatile for different data such as tabular data, images, and time-series. Users only need to collect the input and output data.

• Performant: Because task-based neurons capture the useful prior knowledge from task-related data, the network that is made up of task-based neurons can integrate the task-driven forces, which given the same structure should outperform the network of generic neurons.

Installation

This page provides a brief introduction to graph matching and some guidelines for using pygmtools. If you are seeking some background information, this is the right place!

Important

Please ensure that the versions of packages meet the requirements:

h5py~=3.10.0
numpy~=1.26.2
tnlearn~=0.1
torch~=2.1.0
sympy~=1.12
setuptools~=68.0.0
scikit-learn~=1.4.0
scipy~=1.12.0
joblib~=1.3.2
requests~=2.31.0
networkx~=3.2.1
matplotlib~=3.8.3
pandas~=2.2.0
packaging~=23.2
ipython~=8.18.1
tqdm~=4.66.2

Run the following command to install TNLearn from PyPI:

pip install tnlearn

Documentation

• TNLearn: Task-based Neurons for learning
  • Motivation
  • Features
  • Overview
  • Benchmarks
  • Resource
  • Dependences
  • Install
  • Quick start
  • API documentation
  • Citation
  • The Team
  • License
  • What’s Next?
• Introduction and Guidelines
  • Why Task-based Neurons?
  • What Are Task-based Neurons?
  • When to use Task-based Neurons?
  • What’s Next?
  • Reference
• Get Started
  • Packages Required
  • Basic Install by pip
  • Build Shared Libraries from Source
  • What’s Next
• API Reference
  • tnlearn.VecSymRegressor
  • tnlearn.MLPClassifier
  • tnlearn.MLPRegressor

The Team

tnlearn is a work by:

• Meng Wang (NewT123-WM)

• Fenglei Fan (FengleiFan Fan)

• Juntong Fan (Juntongkuki)

Citing

If you find tnlearn useful, please cite it in your publications.

License

tnlearn is released under the BSD 3-Clause License.

About Version Update

We plan to keep this package up-to-date by including more architectures such as transformer and Mamba.