reservoirpy/reservoirpy

A simple and flexible code for Reservoir Computing architectures like Echo State Networks

What it solves

ReservoirPy is a library designed to simplify the creation and implementation of Reservoir Computing architectures, specifically Echo State Networks (ESN). It provides a flexible framework for handling complex temporal data and time-series prediction, such as chaotic time-series forecasting, and classification tasks.

How it works

The library implements the reservoir computing paradigm where a fixed, random reservoir of neurons is used to map input signals into a high-dimensional space. Only the readout layer is trained, which significantly reduces training time and computation. It supports various configurations, including deep reservoirs, multiple readouts, feedback loops, and sparse matrix computations for efficiency.

Who it’s for

Researchers and developers working with recurrent neural networks, time-series analysis, and dynamical systems who need a flexible tool for designing and optimizing reservoir computing models.

Highlights

  • Flexible Architectures: Supports deep reservoirs, multiple readouts, and feedback connections.
  • Advanced Learning Rules: Includes Intrinsic Plasticity, Local Plasticity, and Next-Generation Reservoir Computing (NVAR).
  • Training Options: Offers both offline and online training modes.
  • Integration: Interfaces with scikit-learn models and provides a Jax backend for acceleration.
  • Optimization: Includes graphical tools for hyperparameter exploration using the hyperopt library.

Related

  • Project
    nschaetti/EchoTorch

    A PyTorch-based module for implementing and testing Echo State Networks and Reservoir Computing, designed for research purposes and integration into deep architectures.

  • Project
    Nixtla/neuralforecast

    A library of over 30 state-of-the-art neural forecasting models providing a user-friendly, scikit-learn style API for high-performance time-series prediction.

  • Project
    skforecast/skforecast

    A Python library for time series forecasting that integrates scikit-learn compatible models, statistical methods, and foundation models into a unified forecasting workflow.

  • Project
    nengo/nengo

    Nengo is a Python library for building and simulating large-scale neural models, supporting both spiking and non-spiking simulations.

  • Project
    google/seqio

    A library for processing sequential data that provides a unified framework for task-based datasets, preprocessing, and evaluation for sequence models.