Abstract. We investigate the properties of ensembles of neural networks, in which each network in the ensemble is constructed using a perturbed version of the training data. The perturbation consists in switching the class labels of a subset of training examples selected at random. Experiments on several UCI and synthetic datasets show that these class-switching ensembles can obtain improvements in classification performance over both individual networks and bagging ensembles. 1

This article investigates the properties of class-switching ensembles composed of neural networks and compares them to class-switching ensembles of decision trees and to standard ensemble learning methods, such as bagging and boosting. In a class-switching ensemble, each learner is constructed using a modified version of the training data. This modification consists in switching the class labels of a fraction of training examples that are selected at random from the original training set. Experiments on 20 benchmark classification problems, including real-world and synthetic data, show that class-switching ensembles composed of neural networks can obtain significant improvements in the generalization accuracy over single neural networks and bagging and boosting ensembles. Furthermore, it is possible to build mediumsized ensembles ( ≈ 200 networks) whose classification performance is comparable to larger class-switching ensembles ( ≈ 1000 learners) of unpruned decision trees.

Abstract. The degree of malignancy in brain glioma needs to be assessed by MRI findings and clinical data before operations. There have been previous attempts to solve this problem by using fuzzy max-min neural networks and support vector machines (SVMs), while in this paper, a novel algorithm named PRIFEB is proposed by combining bagging of SVMs with embedded feature selection for its individuals. PRIFEB is compared with the general case of bagging on UCI data sets, experimental results show PRIFEB can obtain better performance than the general case of bagging. Then, PRIFEB is used to predict the degree of malignancy in brain glioma, computation results show that PRIFEB obtains better accuracy than other several methods like bagging of SVMs and single SVMs does. 1

Abstract. Recently, the core vector machine (CVM) has shown significant speedups on classification and regression problems with massive data sets. Its performance is also almost as accurate as other state-ofthe-art SVM implementations. By incorporating the orthogonality constraints to diversify the CVM ensembles, this turns out to speed up the maximum margin discriminant analysis (MMDA) algorithm. Extensive comparisons with the MMDA ensemble along with bagging on a number of large data sets show that the proposed diversified CVM ensemble can improve classification performance, and is also faster than the original MMDA algorithm by more than an order of magnitude. 1

Abstract—Large-margin methods, such as support vector machines (SVMs), have been very successful in classification problems. Recently, maximum margin discriminant analysis (MMDA) was proposed that extends the large-margin idea to feature extraction. It often outperforms traditional methods such as kernel principal component analysis (KPCA) and kernel Fisher discriminant analysis (KFD). However, as in the SVM, its time complexity is cubic in the number of training points, and is thus computationally inefficient on massive data sets. In this paper, we propose an (1 +) 2-approximation algorithm for obtaining the MMDA features by extending the core vector machine. The resultant time complexity is only linear in, while its space complexity is independent of. Extensive comparisons with the original MMDA, KPCA, and KFD on a number of large data sets show that the proposed feature extractor can improve classification accuracy, and is also faster than these kernel-based methods by over an order of magnitude. Index Terms—Feature extraction, support vector machines (SVMs), core vector machines, scalability.

Using machine learning algorithms

Functional magnetic resonance imaging (fMRI) is becoming a forefront braincomputer interface tool. To decipher brain patterns, fast, accurate and reliable classifier methods are needed. The support vector machines classifier (SVM) has been traditionally used. Here we argue that state-of-the-art methods from pattern recognition and machine learning, such as classifier ensembles, offer more accurate classification. This study compares 18 classification methods on a publicly available real data set due to Haxby and co-authors (2001). The data comes from a single-subject experiment, organised in 10 runs where 8 classes of stimuli were presented in each run. The comparisons were carried out on voxel subsets of different sizes, selected through 7 popular voxel selection methods. We found that, while SVM was robust, accurate and scalable, some classifier ensemble methods demonstrated significantly better performance. The best classifiers were found to be the Random Subspace ensemble of SVM classifiers, Rotation Forest, and ensembles with Random