{"id":13688878,"url":"https://github.com/stefanonardo/pytorch-esn","last_synced_at":"2025-05-01T20:31:05.153Z","repository":{"id":47718433,"uuid":"126745775","full_name":"stefanonardo/pytorch-esn","owner":"stefanonardo","description":"An Echo State Network module for PyTorch.","archived":false,"fork":false,"pushed_at":"2023-02-17T15:07:31.000Z","size":309,"stargazers_count":216,"open_issues_count":6,"forks_count":44,"subscribers_count":9,"default_branch":"master","last_synced_at":"2024-11-12T12:49:27.116Z","etag":null,"topics":["deep-learning","echo-state-networks","esn","machine-learning","neural-networks","python","pytorch","recurrent-neural-networks","reservoir-computing"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/stefanonardo.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2018-03-25T22:28:33.000Z","updated_at":"2024-11-06T11:32:02.000Z","dependencies_parsed_at":"2024-11-12T12:32:54.488Z","dependency_job_id":"db746182-c1fc-416d-b573-a7a933cc691f","html_url":"https://github.com/stefanonardo/pytorch-esn","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/stefanonardo%2Fpytorch-esn","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/stefanonardo%2Fpytorch-esn/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/stefanonardo%2Fpytorch-esn/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/stefanonardo%2Fpytorch-esn/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/stefanonardo","download_url":"https://codeload.github.com/stefanonardo/pytorch-esn/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":251940567,"owners_count":21668558,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","echo-state-networks","esn","machine-learning","neural-networks","python","pytorch","recurrent-neural-networks","reservoir-computing"],"created_at":"2024-08-02T15:01:25.982Z","updated_at":"2025-05-01T20:31:05.148Z","avatar_url":"https://github.com/stefanonardo.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"# PyTorch-ESN\n\nPyTorch-ESN is a PyTorch module, written in Python, implementing Echo State Networks with leaky-integrated units. ESN's implementation with more than one layer is based on [DeepESN](https://arxiv.org/abs/1712.04323). The readout is trainable by ridge regression or by PyTorch's optimizers.\n\nIts development started under my master thesis titled [\"An Empirical Comparison of Recurrent Neural Networks on Sequence Modeling\"](https://www.dropbox.com/s/gyt48dcktht7qur/document.pdf?dl=0), which was supervised by Prof. Alessio Micheli and Dr. Claudio Gallicchio at the University of Pisa.\n\n## Prerequisites\n\n* PyTorch\n\n## Basic Usage\n\n### Offline training (ridge regression)\n\n#### SVD\nMini-batch mode is not allowed with this method.\n\n```python\nfrom torchesn.nn import ESN\nfrom torchesn.utils import prepare_target\n\n# prepare target matrix for offline training\nflat_target = prepare_target(target, seq_lengths, washout)\n\nmodel = ESN(input_size, hidden_size, output_size)\n\n# train\nmodel(input, washout, hidden, flat_target)\n\n# inference\noutput, hidden = model(input, washout, hidden)\n```\n\n#### Cholesky or inverse\n```python\nfrom torchesn.nn import ESN\nfrom torchesn.utils import prepare_target\n\n# prepare target matrix for offline training\nflat_target = prepare_target(target, seq_lengths, washout)\n\nmodel = ESN(input_size, hidden_size, output_size, readout_training='cholesky')\n\n# accumulate matrices for ridge regression\nfor batch in batch_iter:\n    model(batch, washout[batch], hidden, flat_target)\n\n# train\nmodel.fit()\n\n# inference\noutput, hidden = model(input, washout, hidden)\n```\n\n#### Classification tasks\nFor classification, just use one of the previous methods and pass 'mean' or\n'last' to ```output_steps``` argument.\n\n```python\nmodel = ESN(input_size, hidden_size, output_size, output_steps='mean')\n```\n\nFor more information see docstrings or section 4.7 of \"A Practical Guide to Applying\nEcho State Networks\" by Mantas Lukoševičius.\n\n### Online training (PyTorch optimizer)\n\nSame as PyTorch.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fstefanonardo%2Fpytorch-esn","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fstefanonardo%2Fpytorch-esn","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fstefanonardo%2Fpytorch-esn/lists"}