echo-state-network

Python package that provides predictive models for fault detection, soft sensing, and process condition monitoring.

A Python 3 framework for Reservoir Computing with a scikit-learn-compatible API.

Echo State Network (ESN) reservoir computing framework mapping temporal sequences into high-dimensional non-linear dynamical recurrent states.

Echo State Network (ESN) reservoir computing framework mapping temporal sequences into high-dimensional non-linear dynamical recurrent states.

Real-time detection and repair of LLM agent failures — a one-class behavioural monitor at ~200 µs/step, with 2,823 committed traces.

An easy-to-use independent machine learning library for .net. It offers MLP models (including deep RVFL aka ELM) for common ML tasks as well as Reservoir Computer for efficiently solving time series ML tasks.

Implementation of Echo State Networks (ESN) with experiments on MNIST and ECG5000. Includes comparison with Linear Regression and analysis of weight initialization methods for time-series and classification tasks.

Rust crate for chaotic semantic memory - echo state networks and hyperdimensional vectors.

turbESN is an echo state network implementation, used in my PhD research as part of the DeepTurb project of the Carl-Zeiss Stiftung. See https://pypi.org/project/turbESN/

Reservoir computing with coupled genetic oscillators for arrhythmia classification

Julia implementation of cross-scale reservoir computing for high-dimensional spatiotemporal forecasting. Validated on global sea-surface temperature and chaotic dynamics, achieving low error forecasts up to 8 Lyapunov times on Kuramoto-Sivashinsky. Neurocomputing (2026).

Physics-Informed Reservoir Learning for Shallow-Water Flood Modelling

A general-purpose recursive self-improvement loop with a real fly brain: the larval Drosophila connectome as a fixed reservoir, wrapped in a closed, observable L1-L5 loop — search, curricula, gated online adaptation, and an inherited experience bank. The wiring never changes; the lab improves itself. Handwriting is just the first benchmark.

Recursive echo-state reservoir on a 3D Boolean hypercube: each vertex's tanh neuron is replaced by a full sub-reservoir of identical topology, nested to a configurable depth (1-5), with bidirectional parent/child coupling. C++23.

Minimal from-scratch Echo State Network in pure NumPy — and eight experiments showing the rho=0.99 edge-of-chaos rule is wrong in both directions.

rc-bench — воспроизводимая платформа сравнения резервуарных моделей прогнозирования временных рядов: 7 архитектур, честный протокол, реальные данные UCI, ресурсный и энергетический профиль.

Echo State Network (ESN) on a Boolean hypercube reservoir: sparse, XOR-defined connectivity and no stored adjacency. A tunable per-neuron delay line scales memory capacity super-multiplicatively. C++23 and Python SDKs.

Echo state networks for learning variational level set image segmentation. Code behind the MSc dissertation, University of the Witwatersrand, 2018 to 2019.

Reservoir computer on a Boolean hypercube graph. Sparse XOR-defined connectivity. C++23 and Python SDKs.

inZOR-ND: Chaotic systems benchmark — KS_Official structural IID anomaly discovery, +286% improvement (AI-DEEDS 2026)