.png){kind=icon}
Return
PSO-based growing echo state network
.png){kind=icon}
PRE
AI
DOI：10.1016/j.asoc.2019.105774.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Reservoir computing (RC), with the idea of using a large randomly and sparsely connected recurrent layer, has turned out to be an efficient paradigm for training recurrent neural networks. Echo state network (ESN) is one of typical representatives of RC, whose good performance mainly depends on the selection of right parameters. In this paper, we introduce an approach to pre-train a growing ESN with multiple sub-reservoirs by optimizing singular values, based on particle swarm optimization (PSO) and singular value decomposition (SVD). During the network growing, the sub-reservoirs are added to the network one by one, which leads to a quasi-diagonal reservoir weight matrix. For each sub-reservoir, before being added to the network, the singular values are optimized by PSO, then the sub-reservoir weight matrix is designed by SVD with the optimized singular values. In order to evaluate the performance of sub-reservoirs during optimization, a novel fitness function is derived based on some concepts in vector space. Finally, simulation results, on both benchmark and real-world data sets, show the superior performance of the proposed method. (C) 2019 Elsevier B.V. All rights reserved.
Keywords:
Echo state network
Reservoir computing
Particle swarm optimization
Singular value decomposition
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
6.6
Papers:
1.4W
Citations:
4.8W
Organization
Cited Papers
.png){kind=icon}
ShareSave
ShareSaveShareSaveShareSaveShareSaveShareSave
View more