Waveform Classification by Two-Layer Reservoir Computing with van der Pol Oscillators
In this research, one van der Pol oscillator is considered as one node, and the oscillators are interconnected using resistors to form a reservoir layer. A networked reservoir computing model which has two layers is proposed (two-layer reservoir). A waveform classification is performed, and the accuracy is compared with a single-layer reservoir. The results show that changing the nature of the oscillator for each layer improves the classification accuracy. This result suggests that using the two-layer reservoir increases design flexibility and improves accuracy.
Authors
- Yoshifumi Nishio (ORCID: https://orcid.org/0000-0002-0247-0001)
- Yoko Uwate (ORCID: https://orcid.org/0000-0002-2992-8852)
- Yasufumi Kajino
Institutions
- Tokushima University (JP)
Publication Details
- Published
- 2024-08-19
- DOI
- https://doi.org/10.1109/isocc62682.2024.10762273
- Primary Topic
- Neural Networks and Reservoir Computing
- Type
- article
- Field-Weighted Citation Impact
- 0.00