EUSBoost: Enhancing ensembles for highly imbalanced data-sets by evolutionary undersampling

Classification with imbalanced data-sets has become one of the most challenging problems in Data Mining. Being one class much more represented than the other produces undesirable effects in both the learning and classification processes, mainly regarding the minority class. Such a problem needs accurate tools to be undertaken; lately, ensembles of classifiers have emerged as a possible solution. Among ensemble proposals, the combination of Bagging and Boosting with preprocessing techniques has proved its ability to enhance the classification of the minority class.In this paper, we develop a new ensemble construction algorithm (EUSBoost) based on RUSBoost, one of the simplest and most accurate ensemble, which combines random undersampling with Boosting algorithm. Our methodology aims to improve the existing proposals enhancing the performance of the base classifiers by the usage of the evolutionary undersampling approach. Besides, we promote diversity favoring the usage of different subsets of majority class instances to train each base classifier. Centered on two-class highly imbalanced problems, we will prove, supported by the proper statistical analysis, that EUSBoost is able to outperform the state-of-the-art methods based on ensembles. We will also analyze its advantages using kappa-error diagrams, which we adapt to the imbalanced scenario.     

Dynamic classifier selection for One-vs-One strategy: Avoiding non-competent classifiers

The One-vs-One strategy is one of the most commonly used decomposition technique to overcome multi-class classification problems; this way, multi-class problems are divided into easier-to-solve binary classification problems considering pairs of classes from the original problem, which are then learned by independent base classifiers.The way of performing the division produces the so-called non-competence. This problem occurs whenever an instance is classified, since it is submitted to all the base classifiers although the outputs of some of them are not meaningful (they were not trained using the instances from the class of the instance to be classified). This issue may lead to erroneous classifications, because in spite of their incompetence, all classifiers' decisions are usually considered in the aggregation phase.In this paper, we propose a dynamic classifier selection strategy for One-vs-One scheme that tries to avoid the non-competent classifiers when their output is probably not of interest. We consider the neighborhood of each instance to decide whether a classifier may be competent or not. In order to verify the validity of the proposed method, we will carry out a thorough experimental study considering different base classifiers and comparing our proposal with the best performer state-of-the-art aggregation within each base classifier from the five Machine Learning paradigms selected. The findings drawn from the empirical analysis are supported by the appropriate statistical analysis.     

Contrast of a fuzzy relation

In this paper we address a key problem in many fields: how a structured data set can be analyzed in order to take into account the neighborhood of each individual datum. We propose representing the dataset as a fuzzy relation, associating a membership degree with each element of the relation. We then introduce the concept of interval-contrast, a means of aggregating information contained in the immediate neighborhood of each element of the fuzzy relation. The interval-contrast measures the range of membership degrees present in each neighborhood. We use interval-contrasts to define the necessary properties of a contrast measure, construct several different local contrast and total contrast measures that satisfy these properties, and compare our expressions to other definitions of contrast appearing in the literature. Our theoretical results can be applied to several different fields. In an Appendix A, we apply our contrast expressions to photographic images.     

Hydrothermal synthesis and study of an inorganic-organic hybrid vanadate of a nickel(II) coordination complex with pyrazine, Ni3(C4H4N2)3(V8O23)

The three-dimensional hybrid compound Ni₃(C₄H₄N₂)₃(V₈O₂₃) has been synthesized by mild hydrothermal methods under autogenous pressure at 170 °C. The structure of the phase is stable until 380 °C. The removal of the pyrazine molecules from the structure induces its collapse. The IR spectrum shows the vibration modes of the pyrazine molecule and those of the [VO₄]³⁻ groups. A UV-visible spectrum shows the characteristic bands of the Ni(II) d⁸-high-spin cation in a slightly distorted octahedral coordination. Magnetic measurements indicate the existence of antiferromagnetic interactions that can be fitted with a chain model to give g = 2.31, J/k = −5.3, and zJ′/k = −5.5.     

Comment on: “Image thresholding using type II fuzzy sets”. Importance of this method

In this work we develop some reflections on the thresholding algorithm proposed by Tizhoosh in [16]. The purpose of these reflections is to complete the considerations published recently in [17] and [18] on said algorithm. We also prove that under certain constructions, Tizhoosh's algorithm makes it possible to obtain additional information from commonly used fuzzy algorithms.     

Similarity measures,penalty functions,and fuzzy entropy from new fuzzy subsethood measures

In this study, we discuss a new class of fuzzy subsethood measures between fuzzy sets. We propose a new definition of fuzzy subsethood measure as an intersection of other axiomatizations and provide two construction methods to obtain them. The advantage of this new approach is that we can construct fuzzy subsethood measures by aggregating fuzzy implication operators which may satisfy some properties widely studied in literature. We also obtain some of the classical measures such as the one defined by Goguen. The relationships with fuzzy distances, penalty functions, and similarity measures are also investigated. Finally, we provide an illustrative example which makes use of a fuzzy entropy defined by means of our fuzzy subsethood measures for choosing the best fuzzy technique for a specific problem.     

Identification of mannose moieties in N- and O-linked oligosaccharides of the primordial germ cells of Xenopus embryos

The presence of mannose (Man) in the glycoconjugates of primordial germ cells (PGCs) of Xenopus embryos was elucidated by lectin histochemistry with Concanavalin A (Con A) and snowdrop (Galanthus nivalis) bulb lectin (GNA), in combination with deglycosylative pretreatments: beta-elimination, which removes O-linked oligosaccharides, and incubation with Peptide N glycosidase F (PNGase F), which removes N-linked glycan chains. In addition, histochemistry with Con A, which binds to Man and glucose (Glc), was also performed after glucose-oxidase incubation, which converts Glc into gluconic acid, and GNA was carried out after acid hydrolysis, which removes terminal sialic acid (NeuAc) moieties. PGCs were analyzed during their migration over the mesentery until the genital ridge, and after colonization of this gonad anlage. The results showed that for both lectins: (1) the PGCs and other surrounding tissue showed a similar binding pattern, and (2) the staining in the PGCs was similar in the developmental stages studied. Labeling with Con A was due to Man, and not to Glc, as shown after incubation with glucose-oxidase, and it was assumed that Man was in N-linked oligosaccharides. However, GNA labeling was mainly due to O-linked oligosaccharides, because the pretreatment of beta-elimination turned cells negative. Moreover, acid hydrolysis pretreatment gave rise to a stronger GNA-staining, suggesting that either Man was also in subterminal position to NeuAc or some Man-containing glycans were unmasked after removal of NeuAc from other oligosaccharide chains.     

Microstructure and mineralogy of lightweight aggregates produced from washing aggregate sludge,fly ash and used motor oil

The microstructures and mineralogical compositions of lightweight aggregates (LWAs) manufactured with washing aggregate sludge (WS), fly ash (FA) and used motor oil (UMO) have been studied. Most LWAs with WS and FA exhibited an external layer and a glassy core with isolated pores. LWAs with WS and UMO did not present external shells or signs of bloating. Iron oxides, within the external layer, and pyrrhotite, in the inner glass, were observed. The mineralogical analyses revealed the neo-formation of plagioclase and pyroxene, along with minor gehlenite. Some relationships could be established: (i) the presence of larger pores is related to a decrease in the dry particle density values, (ii) when the LWA lacks the external layer, the water absorption values were dependent on the size and amount of each type of pore (open or closed), and (iii) the neo-formation of Ca-plagioclase and the consumption of quartz improved the compressive strength values.     

Construction of interval-valued fuzzy entropy invariant by translations and scalings

In this paper, we propose a method to construct interval-valued fuzzy entropies (Burillo and Bustince 1996). This method uses special aggregation functions applied to interval-contrasts. In this way, we are able to construct interval-valued fuzzy entropies from automorphisms and implication operators. Finally, we study the invariance of our constructions by scaling and translation.     

Definition and construction of fuzzy DI-subsethood measures

In this paper we study a method of construction of the fuzzy subsethood measures of V.R. Young [V.R. Young, Fuzzy subsethood, Fuzzy Sets and Systems 77 (1996) 371-384] and the fuzzy subsethood measures of J. Fan, X. Xie, and J. Pei [J. Fan, X. Xie, J. Pei, Subsethood measures: new definitions, Fuzzy Sets and Systems 106 (1999) 201-209]. We establish the conditions under which our constructions satisfy the axioms of Sinha and Dougherty’s inclusion measures and we present different methods for obtaining fuzzy entropies from said measures. Next we present a particular case of Young’s subsethood measures, DI-subsethood measures. For these we also analyze their construction and the conditions under which they satisfy different axioms.