Search

In this paper, the issue of tailoring the soft confusion matrix (SCM) based classifier to deal with stream learning task is addressed. The main goal of the work is to develop a wrapping-classifier that allows incremental learning to classifiers that are unable to learn incrementally. The goal is achieved by making two improvements in the previously developed SCM classifier. The first one is aimed at reducing the computational cost of the SCM classifier. To do so, the definition of the fuzzy neighborhood of an object is changed. The second one is aimed at effective dealing with the concept drift. This is done by employing the ADWIN-driven concept drift detector that is not only used to detect the drift but also to control the size of the neighbourhood. The obtained experimental results show that the proposed approach significantly outperforms the reference methods.

In this paper, we addressed the problem of building a decision-theoretic classifier tailored for minimizing the Tversky loss under the framework of multi-label classification. The proposed approach is a generalization of the Dembczynski $$F_{\beta }$$ measure optimization algorithm. The introduced technique is based on a series of discrete linear approximations of the Tversky measure. The approximated criterion is then optimized using original optimization algorithm. To assess quality of classification results produced by the designed strategy and compare its outcome with the results obtained by the state-of-the-art approaches, we conducted an experimental study on 24 benchmark datasets. The investigated methods were compared with respect to eleven different quality criteria. We considered quality criteria belonging to three main groups, i.e., example-based, micro-averaged and macro-averaged. During the experimental study, we considered four testing scenarios. Two of them deal with a symmetric variant of the Tversky loss. Remaining scenarios examine asymmetric Tversky loss. The study shows that, in general, the proposed method is comparable to the Dembczynski approach. However, for both symmetric scenarios and one asymmetric scenario, the average ranks suggest that the proposed approach achieves better classification quality in terms of the example-based Tversky measure. This is an important result because the proposed method was designed to optimize the above-mentioned quality indicator. Additionally, the introduced procedure can outperform the reference methods with respect to the zero-one loss under all testing scenarios.

Classifier ensemble pruning is a technique to increase ensemble accuracy and reduce its size. In the paper we propose a novel method of ensemble pruning which is based on the local class and inter-class competences of base classifiers for the test object. In the first step, using the inter-class competences, we identify classifiers that tend to commit the same type of errors. The identified classifiers create a common cluster. In the second step, classifiers in each cluster are ordered according to the class competence and then the top-k ones are selected from each cluster as the final ensemble. Experimental investigations of the proposed pruning method for two homogeneous pools of base classifiers and for three values of parameter $k$ were conducted. In experimental studies we used 16 real data sets related to the various medical decision-making problems. Obtained results clearly show that the use of the ensemble pruning procedure leads to the better classification accuracy.

In this work, we addressed the issue of applying a stochastic classifier and a local, fuzzy confusion matrix under the framework of multi-label classification. We proposed a novel solution to the problem of correcting label pairwise ensembles. The main step of the correction procedure is to compute classifier- specific competence and cross-competence measures, which estimates error pattern of the underlying classifier. We considered two improvements of the method of obtaining confusion matrices. The first one is aimed to deal with imbalanced labels. The other utilizes double labelled instances which are usually removed during the pairwise transformation. The proposed methods were evaluated using 29 benchmark datasets. In order to assess the efficiency of the introduced models, they were compared against 1 state-of-the-art approach and the correction scheme based on the original method of confusion matrix estimation. The comparison was performed using four different multi-label evaluation measures: macro and micro-averaged F1 loss, zero-one loss and Hamming loss. Additionally, we investigated relations between classification quality, which is expressed in terms of different quality criteria, and characteristics of multi-label datasets such as average imbalance ratio or label density. The experimental study reveals that the correction approaches significantly outperforms the reference method only in terms of zero-one loss.

In this paper the problem of recognition of patient’s intent to move hand prosthesis is addressed. The proposed method is based on recognition of electromyographic (EMG) and mechanomyographic (MMG) biosignals using a multiclassifier (MC) system working with dynamic ensemble selection scheme and original concept of competence measure. The concept focuses on developing competence and interclass cross- competence measures which can be applied as a method for classifiers combination. The cross-competence measure allows an ensemble to harness information obtained from incompetent classifiers instead of removing them from the ensemble. The performance of MC system with proposed competence measure was experimentally compared against six state-of-the-art classification methods using real data concerning the recognition of six types of grasping movements. The system developed achieved the highest classification accuracies demonstrating the potential of MC system for the control of bioprosthetic hand.

In the paper the novel feature selection method, using wrapper model and ensemble approach, is presented. In the proposed method features are selected dynamically, i.e. separately for each classified object. First, a set of identical one-feature classifiers using different single feature is created and next the ensemble of features (classifiers) is selected as a solution of optimization problem using genetic algorithm. As an optimality criterion, the sum of measures of features relevance and diversity of ensemble of features is adopted. Both measures are calculated using original concept of randomized reference classifier, which on average acts like classifier with evaluated feature. The performance of the proposed method was compared against six state-of-art feature selection methods using nine benchmark databases. The experimental results clearly show the effectiveness of the dynamic mode and ensemble approach in feature selection procedure.

The paper presents an advanced method of recognition of patient's intention to move hand prosthesis during the grasping and manipulation of objects in a dexterous manner. The proposed method is based on recognition of electromiographic (EMG) and mechanomiographic (MMG) bio signals using two-stage hierarchical multiclassifier system (MCS) with dynamic ensemble selection scheme (DES) and probabilistic competence function. Additionally, the feedback signals derived from the prosthesis sensors are applied to the correction of competences of base classifiers during MCS operation. The performance of proposed MCS was experimetally compared against MCS's without feedback information and with one-stage structure using real data concerning the recognition of five types of grasping movements. The system developed achieved the highest classification accuracy demonstrating the potential of two-stage MCS with feedback signals from prosthesis sensors for the control of bio prosthetic hand.

Dimensionality reduction of attribute set is a common preprocessing step used in machine learning. This step is especially important for high-dimensional data with low-dimensional representation such as gene expression data. Feature reduction is essential in the case of microarray data because most of the microarray data attributes are believed to be unrelated to observed classes. This paper proposes a three-step feature selection framework based on feature clustering, multi-criteria assessment (Borda count) and Markov blanket. The proposed framework is a filter method so it can be used with any classification algorithm. Its classification performance and selection stability were assessed. The experimental studies were performed on 10 microarray data sets. The experimental evaluation showed that the Markov blanket filter produces results comparable to state-of-art methods in terms of classification performance. However it tends to produce unstable solutions.

This paper presents the problem of building the sequential model of the classification task. In our approach the structure of the model is built in the learning phase of classification. In this paper a split criterion based on the MacArthur’s overlapping niches model is proposed. The MacArthur’s overlapping niches distribution is created for each row of the confusion matrix. The split criterion is associated with the analysis of the received distributions. The obtained results were verified on ten data sets. Nine data sets come from UCI repository and one is a real-life data set.