lucapedrelli/deepesn

How lucapedrelli/deepesn works

DeepESN is a Python library for deep reservoir recurrent networks and sequence/time-series tasks. Users construct a model from recurrent layers and a readout, then drive it with a series; task evaluators split data, fit the readout, generate predictions, and report task-specific scores. The graph preserves the documented legacy API, intrinsic-plasticity capability, both task types, dataset utilities, and all readout methods. Some implementation details are unsampled, so edges are limited to documented or indexed relationships.

Main entry points: base.py Readout base and registry · deepesn.py DeepESN model · memory_capacity.py Memory capacity · next_step.py Next-step prediction

Components 19

Model orchestration

DeepESN model src/deepesn/deepesn.py
Legacy configurations src/deepesn/configurations.py

Reservoir layers

Layer base and RNG src/deepesn/layers/base.py
ESN layer src/deepesn/layers/esn.py
Leaky ESN layer src/deepesn/layers/liesn.py

Readout training

Readout base and registry src/deepesn/readouts/base.py
SVD readout src/deepesn/readouts/svd.py
Normal equations src/deepesn/readouts/ne.py
Ridge readout src/deepesn/readouts/ridge.py
Incremental equations src/deepesn/readouts/ine.py
Recursive least squares src/deepesn/readouts/rls.py

Tasks and data

Task evaluation protocol src/deepesn/tasks/base.py
Memory capacity src/deepesn/tasks/memory_capacity.py
Next-step prediction src/deepesn/tasks/next_step.py
Evaluation metrics src/deepesn/metrics.py
Legacy dataset loaders src/deepesn/datasets.py
Legacy task utilities src/deepesn/task.py

Other components

Library user
Predictions and scores

Connections 24

Library user → DeepESN model: constructs README.md
Legacy configurations → DeepESN model: configures legacy API README.md
DeepESN model → Layer base and RNG: initializes layers src/deepesn/deepesn.py
ESN layer → Layer base and RNG: extends src/deepesn/layers/esn.py
Leaky ESN layer → ESN layer: extends src/deepesn/layers/liesn.py
DeepESN model → Readout base and registry: uses readout src/deepesn/deepesn.py
SVD readout → Readout base and registry: implements src/deepesn/readouts/svd.py
Normal equations → Readout base and registry: implements src/deepesn/readouts/ne.py
Ridge readout → Readout base and registry: implements src/deepesn/readouts/ridge.py
Incremental equations → Readout base and registry: implements src/deepesn/readouts/ine.py
Recursive least squares → Readout base and registry: implements src/deepesn/readouts/rls.py
Library user → Memory capacity: evaluates task README.md
Library user → Next-step prediction: evaluates task README.md
Memory capacity → Task evaluation protocol: extends src/deepesn/tasks/memory_capacity.py
Next-step prediction → Task evaluation protocol: extends src/deepesn/tasks/next_step.py
Task evaluation protocol → Evaluation metrics: uses metrics src/deepesn/tasks/base.py
Memory capacity → DeepESN model: computes states src/deepesn/tasks/memory_capacity.py
Memory capacity → Readout base and registry: fits and predicts src/deepesn/tasks/memory_capacity.py
Next-step prediction → DeepESN model: computes states src/deepesn/tasks/next_step.py
Next-step prediction → Readout base and registry: fits and predicts src/deepesn/tasks/next_step.py
Memory capacity → Predictions and scores: returns capacity scores src/deepesn/tasks/memory_capacity.py
Next-step prediction → Predictions and scores: returns predictions and scores src/deepesn/tasks/next_step.py
Legacy task utilities → Legacy dataset loaders: imports utilities src/deepesn/task.py
Library user → Legacy dataset loaders: loads legacy data

Drawn by GitDiagram on November 11, 2026 from the repository's file tree, README and source files. Read as Markdown ·