No abstract

This paper introduces a new modification of the Possibilistic Fuzzy multiclass Novelty Detector for data streams (PFuzzND). Mentioned modification is based on the implementation of the automated adjustment of the number of clusters for each class (determined beforehand or during the novelty detection procedure) to improve algorithm’s ability to divide objects into small groups. As result, the proposed approach generates models with flexible class boundaries, which are capable to identify new classes or extensions of the ones that are already known as well as the outliers. Proposed possibilistic fuzzy algorithm for novelty detection was used to solve various benchmark problems with synthetically generated datasets. In order to show the workability and efficiency of the introduced modification its results were also compared with the results obtained by the original PFuzzND algorithm. Thus, it was established that the PFuzzND technique with automatically adjusted number of clusters allows achieving better results in regards to the accuracy, the Macro F-score metric and the unknown rate measure. Comparison to the original algorithm showed that the proposed modification outperforms it but is sensitive to the parameter settings, which can be also said about the PFuzzND method. Therefore, the MPFuzzND approach can be used instead of the original PFuzzND algorithm for other classification problems.

This study is focused on the automated detection of a complex system operator's condition. For example, in this study a person's reaction while listening to music (or not listening at all) was determined. For this purpose various well-known data mining tools as well as ones developed by authors were used. To be more specific, the following techniques were developed and applied for the mentioned problems: artificial neural networks and fuzzy rule-based classifiers. The neural networks were generated by two modifications of the Differential Evolution algorithm based on the NSGA and MOEA/D schemes, proposed for solving multiobjective optimization problems. Fuzzy logic systems were generated by the population-based algorithm called Co-Operation of Biology Related Algorithms or COBRA. However, firstly each person's state was monitored. Thus, databases for problems described in this study were obtained by using non-contact Doppler sensors. Experimental results demonstrated that automatically generated neural networks and fuzzy rule-based classifiers can properly determine the human condition and reaction. Besides, proposed approaches outperformed alternative data mining tools. However, it was established that fuzzy rule-based classifiers are more accurate and interpretable than neural networks. Thus, they can be used for solving more complex problems related to the automated detection of an operator's condition.