Bio Chem Press  Internet Electronic Journal of Molecular Design is a refereed journal for scientific papers regarding all applications of molecular design
Home | News | Current Issue | Journal Index | IECMD 2004 | Preprint Index | Instructions for Authors | Send the Manuscript | Special Issue
 BioChemPress.com  To bookmark this site press Ctrl D
 
   Home
   News & Announcements
  Journal Info
   Current Issue
   Journal Index
   Preprint Index
   Editor
   Advisory Board
  Conference Info
   IECMD 2004
   Day 1
   Day 2
   Day 3
   Day 4
   Day 5
   Day 6
   Day 7
   Day 8
   Day 9
   Day 10
   IECMD 2003
  BioChem Links
   CoEPrA
   Support Vector Machines
  Author Info
   Instructions for Authors
   Send the Manuscript
   Special Issue
  Contact
   Editorial Office
   Subscription
   Advertising
   Copyright
  User Info
   Terms of Use
   License

Internet Electronic Journal of Molecular Design - IEJMD, ISSN 1538-6414, CODEN IEJMAT
ABSTRACT - Internet Electron. J. Mol. Des. June 2005, Volume 4, Number 6, 381-392

Modeling Structure Property Relationships with Kernel Recursive Least Squares
Rajshekhar, Abhijit Kulkarni, Valadi K. Jayaraman, and Bhaskar D. Kulkarni
Internet Electron. J. Mol. Des. 2005, 4, 381-392

Free: Download the paper in PDF format Return to Table of Contents Get Acrobat Reader to view and print the paper

Abstract:
Modeling structure property relationships accurately is a challenging task and newly developed kernel based methods may provide the accuracy for building these relationships. Kernelized variant of traditional recursive least squares algorithm is used to model two QSPR datasets. All the datasets showed a good correlation between actual and predicted values of boiling points with root mean squared errors (RMSEs) comparable to other conventional methods. For the datasets from Espinosa et al., KRLS showed good prediction statistics with R value in the range of 0.97-0.99 and S value in the range 5.5-8 as compared to multiple linear regression (MLR) with R value in the range 0.85-0.88 and S value in the range 22-26. For the dataset from Trinajstic et al., KRLS performed consistently well with R values lying in the range of 0.95-0.99 and S in the range of 5-10 as compared to MLR with R values in the range of 0.7-0.85 and S in the range of 25-30. The KRLS method works better when more number of variables from the dataset are included as against other methods such as support vector learning or lazy learning technique which works better for smaller number of reduced relevant variables from the dataset.

Free: Download the paper in PDF format Return to Table of Contents Get Acrobat Reader to view and print the paper

Home | News | Current Issue | Journal Index | IECMD 2004 | Preprint Index | Instructions for Authors | Send the Manuscript | Special Issue
Last changes: January 5, 2006 Webmaster
http://www.biochempress.com/
Copyright © 2001-2006 Ovidiu Ivanciuc