Napredna pretraga

Pregled bibliografske jedinice broj: 691475

Development of Gradient Retention Model in Ion Chromatography. Part II: Artificial Intelligence QSRR Approach


Ukić, Šime; Novak, Mirjana; Vlahović, Ana; Avdalović, Nebojša; Liu, Yan; Buszewski, Bogusław; Bolanča, Tomislav
Development of Gradient Retention Model in Ion Chromatography. Part II: Artificial Intelligence QSRR Approach // Chromatographia, 77 (2014), 15; 997-1007 doi:10.1007/s10337-014-2654-4 (međunarodna recenzija, članak, znanstveni)


Naslov
Development of Gradient Retention Model in Ion Chromatography. Part II: Artificial Intelligence QSRR Approach

Autori
Ukić, Šime ; Novak, Mirjana ; Vlahović, Ana ; Avdalović, Nebojša ; Liu, Yan ; Buszewski, Bogusław ; Bolanča, Tomislav

Izvornik
Chromatographia (0009-5893) 77 (2014), 15; 997-1007

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
Ion chromatography; QSRR; artificial intelligence; gradient retention model

Sažetak
Methodology that integrates traditional QSRR modeling with transfer of information from isocratic to gradient environment was presented in the previous paper as efficient new chromatographic approach. The previous research included application of conventional regression techniques what resulted with relatively high prediction errors. Also, it was shown that prediction error of the integrated model was mostly caused by prediction of the QSRR models. Therefore, artificial intelligence was applied in this work in order to improve the prediction ability of QSRR-based model: Artificial neural networks were selected for QSRR modeling, while genetic algorithm was used for the selection of optimal descriptors. Both artificial neural networks and genetic algorithm were optimized in order to build an accurate and reliable QSRR models. Selection function, crossover function, and percentage of genes’ mutations were varied in case of genetic algorithm, while artificial neural networks were optimized by means of different network type, training algorithm and number of neurons in hidden layer. During retention modeling, basic QSRR models developed for specific eluent strength were upgraded to isocratic, and thereafter to gradient retention model. None of three developed models showed systematic error, and the obtained predictions (RMSEP 11.66, 10.67, and 7.10 %, respectively) indicated significant improvement from the results presented in previous paper.

Izvorni jezik
Engleski

Znanstvena područja
Kemija, Kemijsko inženjerstvo



POVEZANOST RADA


Projekt / tema
110005
125-1253092-3004 - Procesi ionske izmjene u sustavu kvalitete industrijskih voda (Tomislav Bolanča, )

Ustanove
Fakultet kemijskog inženjerstva i tehnologije, Zagreb

Časopis indeksira:


  • Current Contents Connect (CCC)
  • Web of Science Core Collection (WoSCC)
    • Science Citation Index Expanded (SCI-EXP)
    • SCI-EXP, SSCI i/ili A&HCI
  • Scopus


Citati