ACIIDS'11 Proceedings of the Third international conference on Intelligent information and database systems - Volume Part I
Using rotation forest for protein fold prediction problem: an empirical study
EvoBIO'10 Proceedings of the 8th European conference on Evolutionary Computation, Machine Learning and Data Mining in Bioinformatics
Protein fold recognition with a two-layer method based on SVM-SA, WP-NN and C4.5 TLM-SNC
International Journal of Data Mining and Bioinformatics
FRAN and RBF-PSO as two components of a hyper framework to recognize protein folds
Computers in Biology and Medicine
Comparing ensemble learning methods based on decision tree classifiers for protein fold recognition
International Journal of Data Mining and Bioinformatics
Hi-index | 0.00 |
Protein fold pattern recognition has been one of the most challenging problems in biology during the last 40 years. Recently due to the vast improvement in machine learning and pattern recognition methods many computer scientists have applied these methods to solve this problem. However, protein folding problem is much more complicated than ordinary machine learning problems because of its natural complexity imposed by the high dimensionality of feature space and diversity of different protein fold classes. To deal with such a challenging problem, we use an ensemble classifier model by applying MLP and RBF Neural Networks and Bayesian ensemble method. Also we have used the Laplace estimation method in order to smooth confusion matrices of the base classifiers. Experimental results imply that RBF Neural Network holds better Correct Classification Rate (CCR) compared to other common classification methods such as MLP networks. Our experiments also show that the Bayesian fusion method can improve the correct classification rate of proteins up to 20% with the final CCR of 59% by reducing both bias and variance error of the RBF classifiers, on a benchmark dataset containing 27 SCOP folds.