Search PubMed⌕ Search
Home/Archive/PubMed

PubMed · 16013311

[Improving partial least square regression precision in NIR multi-component analysis using artificial neural network].

Ying-kui Bai·Xian-jiang Meng·Dong Ding·Xuan-guo Shen
View original source ↗Export metadata ↓

Abstract

The present paper presents a new NIR multi-component analysis method with Artificial Neural Network(ANN) and Partial Least Square Regression(PLS). First, this method divides the concentration range of training samples into some sub-ranges, and respectively computes a PLS correlation model in each sub-range with the sub-range's training samples. Then, the authors classify prediction samples according to its concentration sub-range with ANN and judge which sub-range theprediction sample belongs to. Finally, the authors compute the concentration of prediction component with the PLS correlation model of the sub-range according to ANN. The experiment and the result of data processing show that this method improves the model's applicability, and evidently enhances prediction precision compared to traditional PLS.

Explore related subjects

Least-Squares AnalysisNeural Networks, ComputerPharmaceutical PreparationsReproducibility of ResultsSpectroscopy, Near-InfraredTechnology, PharmaceuticalJournal Article

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Ying-kui Bai, Xian-jiang Meng, Dong Ding, Xuan-guo Shen. 2005. [Improving partial least square regression precision in NIR multi-component analysis using artificial neural network].. https://pubmed.ncbi.nlm.nih.gov/16013311/

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2006

Quantitative structure-property relationships for direct photolysis of polybrominated diphenyl ethers.

Using semiempirical quantum chemical descriptors, by partial least squares (PLS) regression, quantitative structure-property relationships (QSPRs) were established for direct photolysis quantum yields (Phi) and rate constants (k) of polybrominated diphenyl ether congeners dissolved in water/methanol and methanol solutions, respectively, and irradiated by artificial ultraviolet A light. Q(cum)(2), a parameter indicating robustness and predictive abilities of PLS models, for the significant QSPR models is larger than 0.702. The gap of frontier molecular orbital energies (E(LUMO)-E(HOMO)) and the most positive Mulliken atomic charges on a hydrogen atom (q(H+)) are two main molecular structural factors governing the logPhi values. logPhi increases with increasing E(LUMO)-E(HOMO) and q(H+) values. logk is mainly related to bromination degree and pattern which can be characterized by molecular weight (Mw), average molecular polarizability (alpha), and average Mulliken atomic charges on bromine atoms (q(Br)). logk increases with bromination degree (Mw, alpha) and q(Br).

Least-Squares Analysis↗
PubMed2006

Application of spectral beta-correction method and partial least squares for simultaneous determination of V(IV) and V(V) in surfactant media.

Simultaneous determination of V(IV) and V(V) was performed by application of partial least squares when the calibration matrix was obtained using beta-correction spectra. Two reaction between V(IV) and V(V) and Pyrogallol Red as a ligand in presence of cethyltrimethylammoniumbromide (CTAB) media has been investigated and applied to the simultaneous spectrophotometric determination of these species. The parameters controlling behavior of the system were investigated and optimum conditions selected. Determinations were made over the concentration range 0.6-4.50 microg ml-1 of V(IV) and 0.3-5.50 microg ml-1 of V(V). Applying this method to simultaneous determination of these metal ions in several real samples with total relative standard error less than 5% validated the proposed method.

Least-Squares Analysis↗
PubMed2005

Prediction of the tissue/blood partition coefficients of organic compounds based on the molecular structure using least-squares support vector machines.

The accurate nonlinear model for predicting the tissue/blood partition coefficients (PC) of organic compounds in different tissues was firstly developed based on least-squares support vector machines (LS-SVM), as a novel machine learning technique, by using the compounds' molecular descriptors calculated from the structure alone and the composition features of tissues. The heuristic method (HM) was used to select the appropriate molecular descriptors and build the linear model. The prediction result of the LS-SVM model is much better than that obtained by HM method and the prediction values of tissue/blood partition coefficients based on the LS-SVM model are in good agreement with the experimental values, which proved that nonlinear model can simulate the relationship between the structural descriptors, the tissue composition and the tissue/blood partition coefficients more accurately as well as LS-SVM was a powerful and promising tool in the prediction of the tissue/blood partition behaviour of compounds. Furthermore, this paper provided a new and effective method for predicting the tissue/blood partition behaviour of the compounds in the different tissues from their structures and gave some insight into structural features related to the partition process of the organic compounds in different tissues.

Least-Squares Analysis↗