Artificial neural networks as an alternative to the traditional statistical methodology in plant research
Artificial neural networks as an alternative to the traditional statistical methodology in plant research
复制标题
DOI：
10.1016/j.jplph.2009.07.007
复制
发表时间：
2010-01-01
影响因子：
4.3
通讯作者：
Gallego, P. P.
中科院分区：
文献类型：
--
作者：
Gago, J.;Martinez-Nunez, L.;Gallego, P. P.
在这项工作中，我们比较了独特的人工神经网络(ANN)技术和通常的统计分析，以确定其作为一种替代方法在植物研究中的实用性。为此，我们选择了一个简单的体外增殖实验，目的是评估光照强度和蔗糖浓度对外植体增殖成功的影响，并在考虑所有影响因素的情况下优化工艺。通过数据分析，传统的统计方法和人工神经网络技术都表明，在本试验条件下，猕猴桃试管苗的最高增殖率需要弱光处理和高蔗糖浓度。然而，这种特殊的人工神经网络软件能够对过程进行建模和优化，以估计最佳条件，并且不需要非常专业的背景。讨论了神经网络方法在分析植物生物学过程中的潜力，在这种情况下，植物组织培养数据。(C)2009年爱思唯尔股份有限公司。版权所有。
In this work, we compared the unique artificial neural networks (ANNs) technology with the usual statistical analysis to establish its utility as an alternative methodology in plant research. For this purpose, we selected a simple in vitro proliferation experiment with the aim of evaluating the effects of light intensity and sucrose concentration on the success of the explant proliferation and finally, of optimizing the process taking into account any influencing factors. After data analysis, the traditional statistical procedure and ANNs technology both indicated that low light treatments and high sucrose concentrations are required for the highest kiwifruit microshoot proliferation under experimental conditions. However, this particular ANNs software is able to model and optimize the process to estimate the best conditions and does not need an extremely specialized background. The potential of the ANNs approach for analyzing plant biology processes, in this case, plant tissue culture data, is discussed. (C) 2009 Elsevier GmbH. All rights reserved.