Reservoirpy

Major changes

- Experimental: Leaky-integrate and fire spiking reservoir.
This is a first step towards a complete integration of Liquid State Machines into ReservoirPy. Please send us feedback on this experimental node by mail or by raising an issue !
- Ring topology: neuron connections forms a loop
- Orthogonal recurrent matrices
- Line topology: neuron connections forms a line
- Consistency: Existing params and hypers can be modified directly (`node.param = ...` instead of `node.params["param"]`)
- Observables:
- memory capacity
- Dimensionwise scores: observables can now return a list of scores for each dimension, with the dimensionwise parameter.
- Effective spectral radius: metrics proposed in [(Jaeger, 2007)](https://doi.org/10.1016/j.neunet.2007.04.016)
- Dataset:
- Multiple Superimposed Oscillator
- Santa-Fe laser dataset (fixes 171 )
- Easy import of classification datasets from the timeseriesclassification.com website
- One-hot encoder that is ReservoirPy-compatible


Minor changes

- NumPy v2.0 compatibility (fixes 169 )
- Bugfix: ESN node with feedback failed (bug introduced in latest version) (fixes 166 )
- Bugfix: Weird behavior of hyperparameter plotting when one of the parameters had a linear scale (fixes 106 )
- Bugfix: hyperparameter searches with "randint" hyperparameter scale failed to be saved correctly in the json file
- Bugfix: Incorrect model creation with many node-to-node connections
- JSON files generated with hyperparameter exploration are now pretty-printed
- Some error messages improvements & documentation correction (fixes 167 , fixes 160 )

Major changes

- Fixed `numpy.memmap` concurrent accesses. Which fixes many issues with parallelization. The memmap name is now unique for each buffer.
Fixes 141, fixes 112, fixes 57
- Sparse matrix initializers in `mat_gen` can now take a `degree` argument instead of `connectivity`.
- `Delay` node, which returns its input with a specified delay.
- Complete reimplementation of the `ScikitLearnNode` introduced in ReservoirPy v0.3.10. This has breaking changes, as the previous implementation was not consistent with the library node API. The tutorial has subsequently been updated.
- Dropped support for Python 3.6. Python 3.7 should still work though it reached [end-of-life](https://devguide.python.org/versions/).

Minor changes

- `dataset.narma` now takes a `u` parameter so that the input timeseries can be accessed. Fixes 142
- Fix : The ESN node now has consistent results across backends
- Creating a Reservoir node with incorrect argument now raises a `ValueError`. Fixes 138
- Fix: the default `Node.partial_backward` method of offline nodes didn't concatenate output timeseries correctly.
- Many small corrections in documentation. In particular:
- Fix 134
- Fix 111
- Added documentation for the `Concat` node

Major changes

- Addition of a ScikitLearnNode : [scikit-learn linear models](https://scikit-learn.org/stable/modules/linear_model.html) can now be used as ReservoirPy's nodes. This can be used in particular for classification.
A detailed tutorial on how to use this new node can be found [here](https://github.com/reservoirpy/reservoirpy/blob/master/tutorials/6-Sklearn-node.ipynb).
This fixes 82

(by Deepayan137 )

Minor changes

- Removed SciPy upper bound, as older scipy versions are now unsupported and uncompatible with more recent Python versions
This fixes 128 and 135
- Fix: synchronized dependencies versions across the different config files

Major Fix

- `ESN.run` had the same problem as `ESN.fit` when using `multiprocessing` as joblib backend.

**Full Changelog**: https://github.com/reservoirpy/reservoirpy/compare/v0.3.9...v0.3.9.post1

Major fixes

- Deterministic behavior of Concat node: fix a major issue with Concat nodes being fed inputs in random order (115 114 by PAUL-BERNARD)
- `japanese_vowels` link was dead, the dataset could not be loaded (113).
- Using `multiprocessing` as joblib backend was failing because of a locale function in `ESN.fit`.

Minor Fixes

- Fixed rsquare averaging y_pred instead of y_true (110 by HugoChateauLaurent).
- Improved robustness of spectral radius computation for reproducibility (116 by PAUL-BERNARD) .
- `dataset` module coherence and precision: Add `**kwargs` everywhere, fix some issues with timeseries length (118 117)
- Typos in Reservoir equations (101)

New Contributors
* PAUL-BERNARD made their first contribution in https://github.com/reservoirpy/reservoirpy/pull/116
* HugoChateauLaurent made their first contribution in https://github.com/reservoirpy/reservoirpy/pull/110

**Full Changelog**: https://github.com/reservoirpy/reservoirpy/compare/v0.3.8...v0.3.9

Major changes

- Leak rate `lr` in reservoir nodes can now be an array or list of float values, one per neurons.

What's Changed
* V0.3.8 by nTrouvain in https://github.com/reservoirpy/reservoirpy/pull/108


**Full Changelog**: https://github.com/reservoirpy/reservoirpy/compare/v0.3.7...v0.3.8