EchoBay is a C++, Eigen powered library, for the training and deployment of Echo State Networks, with multiple layers and different topologies. EchoBay employs Limbo library to find the best set of hyper-parameters that maximize a score function. EchoBay is designed to work almost seamlessly on small computers, microcontrollers (tested on ESP32) and large multi-threaded systems.
NOTE This repo is read-only now for reference purposes of manuscript https://doi.org/10.1145/3404993. Current active development and updated dependencies are available at: https://github.com/LucaCerina/Echobay
Optional dependencies to read CSV data fast-cpp-csv-parser.
NOTE This repo is read-only and prerequisites may not be updated. Current active development and updated dependencies are available at: https://github.com/LucaCerina/Echobay
Full documentation is available here.
After configuration and building (see docs for details) the basic usage is:
./echobay train configuration.yml outputFolderName
./echobay compute modelFolderName
Other important contributions in this research are from Claudio Gallicchio and Alessio Micheli (Department of Computer Science, University of Pisa) and Marco D. Santambrogio (Dipartimento di Informatica, Elettronica e Bioingegneria, Politecnico di Milano)
If you use EchoBay in a scientific paper, please cite:
L. Cerina, G. Franco, C. Gallicchio, and A. Micheli, and M. D. Santambrogio. 2020. EchoBay: Design and Optimization of Echo State Networks under Memory and Time Constraints. ACM Trans. Archit. Code Optim. 17, 3, Article 22 (September 2020), 24 pages. https://doi.org/10.1145/3404993