An introduction to support Vector Machines: and other kernel-based learning methods
An introduction to support Vector Machines: and other kernel-based learning methods
Two-stage support vector machines for protein secondary structure prediction
Neural, Parallel & Scientific Computations - Special issue: Advances in intelligent systems and applications
Estimation of Dependences Based on Empirical Data: Springer Series in Statistics (Springer Series in Statistics)
A comparison of methods for multiclass support vector machines
IEEE Transactions on Neural Networks
Di-codon Usage for Gene Classification
PRIB '09 Proceedings of the 4th IAPR International Conference on Pattern Recognition in Bioinformatics
EvoBIO'07 Proceedings of the 5th European conference on Evolutionary computation, machine learning and data mining in bioinformatics
ICIC'09 Proceedings of the Intelligent computing 5th international conference on Emerging intelligent computing technology and applications
Identification of true EST alignments for recognising transcribed regions
International Journal of Data Mining and Bioinformatics
International Journal of Data Mining and Bioinformatics
Support Vector Machines with L1 penalty for detecting gene-gene interactions
International Journal of Data Mining and Bioinformatics
Prediction of protein secondary structure using large margin nearest neighbour classification
International Journal of Bioinformatics Research and Applications
International Journal of Data Mining and Bioinformatics
A sampling approach for protein backbone fragment conformations
International Journal of Data Mining and Bioinformatics
Hi-index | 0.00 |
Bioinformatics techniques to Protein Secondary Structure (PSS)prediction mostly depend on the information available in amino acidsequences. In this paper, we propose a two-stage Multi-classSupport Vector Machine (MSVM) approach, where the second MSVMpredictor is introduced at the output of the first stage MSVM tocapture the contextual relationship among secondary structureelements in order to minimise the generalisation error in theprediction. By using position-specific scoring matrices generatedby PSI-BLAST, the two-stage MSVM approach achieves Q3accuracies of 78.0% and 76.3% on the RS126 dataset of 126non-homologous globular proteins and the CB396 dataset of 396non-homologous proteins, respectively, which are better than thescores reported on both datasets to date. By using MSVM, thepresent prediction scheme significantly achieves 2 6% and 3 15% ofimprovement in Q3 and Sov accuracies, respectively, onthe two datasets. On larger blind-test datasets from PSIPRED, CASP4and EVA datasets, two-stage MSVM approach achieves Q3accuracies from 77.0% to 79.5%.