Spectral peak verification and recognition using a multilayered neural network.
Spectral peak verification and recognition using a multilayered neural network.
复制标题
使用多层神经网络进行光谱峰值验证和识别。
DOI：
10.1021/ac00223a011
复制
发表时间：
1990
影响因子：
7.4
通讯作者：
Tomellini,SA
中科院分区：
文献类型：
--
作者：
Wythoff,BJ;Levine,SP;Tomellini,SA
引言人工神经网络是生物神经系统的数学模型。尽管它们是对实际物理认知过程的粗略简化，但这些模型的应用表明人工神经网络在与人类相同的领域具有优点和缺点。例如，它们擅长识别视觉模式，但不太适合精确的数学计算。尽管模式识别的传统数学方法取得了进步，但人类通常仍然被认为是可用于听觉和视觉模式的最有效的模式识别系统。总之，人类在视觉模式识别方面的优势以及生物和人工神经网络之间的联系为即将描述的研究提供了灵感。本研究的目的是研究人工神经网络再现人类对红外光谱数据中峰形信号是否存在的判断的能力。
INTRODUCTION Artificial neural networks are mathematical models of biological neural systems. Although they are a gross simplifi-cation of actual physical cognitive processes, application of these models has indicated that artificial neural networks have strengths andweaknesses in the same areas as humans. For example, they excel at recognition of visual patterns but are poorly suited to precise mathematical calculation. Despite the advances in traditional mathematical approaches to pattern recognition, humans are generally still considered to be the most effective pattern recognition system available for audible and visual patterns. Together, the superiority of humans at visual pattern recognition and the link between biological and artificial neural networks provided the inspi-ration for the research to be described. The object of this research was to investigate the ability of an artificial neural network to reproduce human judgements on the presence of peak-shaped signals in infrared spectral data.