Mod-DeepESN: Modular Deep Echo State Network
Machine Learning
2019-03-27 v2 Machine Learning
Abstract
Neuro-inspired recurrent neural network algorithms, such as echo state networks, are computationally lightweight and thereby map well onto untethered devices. The baseline echo state network algorithms are shown to be efficient in solving small-scale spatio-temporal problems. However, they underperform for complex tasks that are characterized by multi-scale structures. In this research, an intrinsic plasticity-infused modular deep echo state network architecture is proposed to solve complex and multiple timescale temporal tasks. It outperforms state-of-the-art for time series prediction tasks.
Cite
@article{arxiv.1808.00523,
title = {Mod-DeepESN: Modular Deep Echo State Network},
author = {Zachariah Carmichael and Humza Syed and Stuart Burtner and Dhireesha Kudithipudi},
journal= {arXiv preprint arXiv:1808.00523},
year = {2019}
}
Comments
4 pages, Cognitive Computational Neuroscience (CCN) 2018 Conference