A hybrid approach to learn with imbalanced classes using evolutionary algorithms

There is an increasing interest in the application of Evolutionary Algorithms (EAs) to induce classification rules. This hybrid approach can benefit areas where classical methods for rule induction have not been very successful. One example is the induction of classification rules in imbalanced domains. Imbalanced data occur when one or more classes heavily outnumber other classes. Frequently, classical machine learning (ML) classifiers are not able to learn in the presence of imbalanced data sets, inducing classification models that always predict the most numerous classes. In this work, we propose a novel hybrid approach to deal with this problem. We create several balanced data sets with all minority class cases and a random sample of majority class cases. These balanced data sets are fed to classical ML systems that produce rule sets. The rule sets are combined creating a pool of rules and an EA is used to build a classifier from this pool of rules. This hybrid approach has some advantages over undersampling, since it reduces the amount of discarded information, and some advantages over oversampling, since it avoids overfitting. The proposed approach was experimentally analysed and the experimental results show an improvement in the classification performance measured as the area under the receiver operating characteristics (ROC) curve.

Two-level network design with intermediate facilities: An application to electrical distribution systems

We consider the two-level network design problem with intermediate facilities. This problem consists of designing a minimum cost network respecting some requirements, usually described in terms of the network topology or in terms of a desired flow of commodities between source and destination vertices. Each selected link must receive one of two types of edge facilities and the connection of different edge facilities requires a costly and capacitated vertex facility. We propose a hybrid decomposition approach which heuristically obtains tentative solutions for the vertex facilities number and location and use these solutions to limit the computational burden of a branch-and-cut algorithm. We test our method on instances of the power system secondary distribution network design problem. The results show that the method is efficient both in terms of solution quality and computational times. (C) 2010 Elsevier Ltd. All rights reserved.

Carbon nanotubes in nanostructured films: Potential application as amperometric and potentiometric field-effect (bio-)chemical sensors

The assembly of carbon nanotubes (CNTs) into nanostructured films is attractive for producing functionalized hybrid materials and (bio-)chemical sensors, but this requires experimental methods that allow for control of molecular architecturcs. In this study, we exploit the layer-by-layer (LbL) technique to obtain two types of sensors incorporating CNTs. In the first, LbL films of alternating layers of multi-walled carbon nanotubes (MWNTs) dispersed in polyarninoamide (PAMAM) dendrimers and nickel phthalocyanine (NiTsPc) were used in amperometric detection of the neurotransmitter dopamine (DA). The electrochemical properties evaluated with cyclic voltammetry indicated that the incorporation of MWNTs in the PAMAM-NT/NiTsPc LbL films led to a 3-fold increase in the peak current, in addition to a decrease of 50 mV in the oxidation potential of DA. The latter allowed detection of DA even in the presence of ascorbic acid (AA), a typical interferent for DA. Another LbL film was obtained with layers of PAMAM and single-walled carbon nanotubes (SWNTs) employed in field-effect-devices using a capacitive electrolyte-insulator-semiconductor structure (EIS). The adsorption of the film components was monitored by measuring the flat-band voltage shift in capacitance-voltage (C-P) curves, caused by the charges from the components. Constant capacitance (ConCap) measurements showed that the EISPAMAM/SWNT film displayed a high pH sensitivity (ca. 54.5 mV/pH), being capable of detecting penicillin G between 10(-4) mol L(-1) and 10(-2) mol L-1, when a layer of penicillinase was adsorbed atop the PAMAM/SWNT film. (C) 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.