【摘要】：波長選擇已經(jīng)成為了分子光譜模型優(yōu)化的一個重要環(huán)節(jié)。在大多數(shù)情況下,我們只關(guān)心波長選擇后模型預(yù)測效果能夠提高多少,而波長選擇算法在實(shí)際應(yīng)用中存在的一些問題往往會被忽略。本文總結(jié)了這些問題和對應(yīng)的一些解決方案或思路。例如不同光譜預(yù)處理方法對變量選擇算法的應(yīng)用效果影響如何,部分常用波長選擇算法選擇變量的穩(wěn)定性和可靠性如何提高,以及不同類型分子光譜數(shù)據(jù)對波長選擇算法的應(yīng)用效果影響如何等。總之,我們希望以上這些問題的提出不僅能夠引起廣大分子光譜同行的重視,進(jìn)而促進(jìn)波長選擇算法在分子光譜領(lǐng)域取得更好的應(yīng)用效果。
[Abstract]:Wavelength selection has become an important part of molecular spectral model optimization. In most cases, we only care how much the model prediction effect can be improved after wavelength selection, but some problems in practical application of wavelength selection algorithm are often ignored. This paper summarizes these problems and some corresponding solutions or ideas. For example, the influence of different spectral preprocessing methods on the application effect of variable selection algorithm, and how to improve the stability and reliability of some commonly used wavelength selection algorithms, And the influence of different kinds of molecular spectral data on the application effect of wavelength selection algorithm and so on. In a word, we hope that these problems can not only arouse the attention of the broad spectrum peers, but also promote the better application of wavelength selection algorithm in the field of molecular spectrum.
【作者單位】： 中國農(nóng)業(yè)大學(xué)理學(xué)院;
【分類號】：O657.3

【相似文獻(xiàn)】

相關(guān)期刊論文 前2條

1 洪明堅;溫志渝;;一種多模型融合的近紅外波長選擇算法[J];光譜學(xué)與光譜分析;2010年08期

2 ;[J];;年期

，

本文編號：2183195