期刊
JOURNAL OF APPLIED PHYSICS
卷 124, 期 16, 页码 -出版社
AIP Publishing
DOI: 10.1063/1.5039826
关键词
-
资金
- Labex ACTION program [ANR-11-LABX-0001-01]
- BiPhoProc ANR Project [ANR-14-OHRI-0002-02]
- Agence Nationale de la Recherche (ANR) [ANR-14-OHRI-0002] Funding Source: Agence Nationale de la Recherche (ANR)
In this work, we propose a new approach toward the efficient optimization and implementation of reservoir computing hardware, reducing the required domain-expert knowledge and optimization effort. First, we introduce a self-adapting reservoir input mask to the structure of the data via linear autoencoders. We, therefore, incorporate the advantages of dimensionality reduction and dimensionality expansion achieved by conventional algorithmically-efficient linear algebra procedures of principal component analysis. Second, we employ evolutionary-inspired genetic algorithm techniques resulting in a highly efficient optimization of reservoir dynamics with a dramatically reduced number of evaluations comparing to exhaustive search. We illustrate the method on the so-called single-node reservoir computing architecture, especially suitable for implementation in ultrahighspeed hardware. The combination of both methods and the resulting reduction of time required for performance optimization of a hardware system establish a strategy toward machine learning hardware capable of self-adaption to optimally solve specific problems. We confirm the validity of those principles building reservoir computing hardware based on a field-programmable gate array. Published by AIP Publishing.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据