DeepESN2019a: Deep Echo State Network (DeepESN) Toolbox v1.1
1,5K download
Aggiornato 7 feb 2019

Visualizza la licenza

DeepESN2019a - Deep Echo State Network Toolbox v1.1 (February 2019)

Deep Echo State Networks (DeepESN) extend the Reservoir Computing paradigm towards the Deep Learning framework.
Essentially, a DeepESN is a deep Recurrent Neural Network composed of a stacked composition of multiple recurrent reservoir layers, and of a linear readout layer that computes the output of the model. The deep reservoir part is left untrained after initialization, and the readout is the only part of the architecture that undergoes a training process.
All details on the DeepESN model can be found in the reference paper reported below in the CITATION REQUEST section.
Also note that DeepESNs with a single layer reduce to standard Echo State Networks (ESNs), thereby the code provided in this toolbox can also be used for standard (i.e., shallow) ESN applications.

The toolbox contains the files listed below.
- DeepESN.m: The file contains the definition of the class DeepESN (the main class in the toolbox).
- Task.m: The file contains the definition of the auxiliary class Task.
- example_DeepESN_1.m: The file contains an example of the usage of the classes in the DeepESN toolbox for the short-term Memory Capacity (MC) task.
- example_task_MC.m: The file contains an example of the usage of the methods in the Task class, including loading of (input and target) data from .csv files, and hold-out cross-validation settings.
- MC100.mat: The file contains an object of class Task, representing the information for the MC task (up to 100 reservoir units), used in the provided example code. This file contains the Task object obtained by running example_task_MC.m
- MC_input.csv, MC_target.csv: files in csv format containing the input and target data for the MC task.

All the files come with full documentation, accessible through the individual reference pages, or through the help function. E.g., for info on the DeepESN class, type 'help DeepESN' in the Matlab command window.

The DeepESN model has been proposed in the following journal paper, which represents a citation request for the usage of this toolbox:
C. Gallicchio, A. Micheli, L. Pedrelli, "Deep Reservoir Computing: A Critical Experimental Analysis", Neurocomputing, 2017, vol. 268, pp. 87-99

An up-to-date overview of the research developments on DeepESN can be found in:
C. Gallicchio, A. Micheli, "Deep Echo State Network (DeepESN): A brief survey", arXiv preprint arXiv:171204323, 2018

Claudio Gallicchio -

Department of Computer Science - University of Pisa (Italy)
Computational Intelligence & Machine Learning (CIML) Group

Cita come

Gallicchio, Claudio, et al. “Deep Reservoir Computing: A Critical Experimental Analysis.” Neurocomputing, vol. 268, Elsevier BV, Dec. 2017, pp. 87–99, doi:10.1016/j.neucom.2016.12.089.

Visualizza più stili
Compatibilità della release di MATLAB
Creato con R2018b
Compatibile con qualsiasi release
Compatibilità della piattaforma
Windows macOS Linux
Scopri di più su Sequence and Numeric Feature Data Workflows in Help Center e MATLAB Answers

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!
Versione Pubblicato Note della release

- fixed minor issue with inter-layer initialization
- added example file for the setting of a Task object (along with .csv files for input/target data)


Updated description.