Eduardo Corrêa Gonçalves

A Genetic Algorithm for Optimizing the Label Ordering in Multi-label Classifier Chains

First proposed in 2009, the classifier chains model (CC) has become one of the most influential algorithms for multi-label classification. It is distinguished by its simple and effective approach to exploit label dependencies. The CC method involves the training of q single-label binary classifiers, where each one is solely responsible for classifying a specific label in ll, ..., lq. These q classifiers are linked in a chain, such that each binary classifier is able to consider the labels predicted by the previous ones as additional information at classification time. The label ordering has a strong effect on predictive accuracy, however it is decided at random and/or combining random orders via an ensemble. A disadvantage of the ensemble approach consists of the fact that it is not suitable when the goal is to generate interpretable classifiers. To tackle this problem, in this work we propose a genetic algorithm for optimizing the label ordering in classifier chains. Experiments on diverse benchmark datasets, followed by the Wilcoxon test for assessing statistical significance, indicate that the proposed strategy produces more accurate classifiers.

Distinct chains for different instances: an effective strategy for multi-label classifier chains

Multi-label classification (MLC) is a predictive problem in which an object may be associated with multiple labels. One of the most prominent MLC methods is the classifier chains (CC). This method induces q binary classifiers, where q represents the number of labels. Each one is responsible for predicting a specific label. These q classifiers are linked in a chain, such that at classification time each classifier considers the labels predicted by the previous ones as additional information. Although the performance of CC is largely influenced by the chain ordering, the original method uses a random ordering. To cope with this problem, in this paper we propose a novel method which is capable of finding a specific and more effective chain for each new instance to be classified. Experiments have shown that the proposed method obtained, overall, higher predictive accuracies than the well-established binary relevance, CC and CC ensemble methods.

Simpler is Better: a Novel Genetic Algorithm to Induce Compact Multi-label Chain Classifiers

Multi-label classification (MLC) is the task of assigning multiple class labels to an object based on the features that describe the object. One of the most effective MLC methods is known as Classifier Chains (CC). This approach consists in training q binary classifiers linked in a chain, y1 → y2 → ... → yq, with each responsible for classifying a specific label in {l1, l2, ..., lq}. The chaining mechanism allows each individual classifier to incorporate the predictions of the previous ones as additional information at classification time. Thus, possible correlations among labels can be automatically exploited. Nevertheless, CC suffers from two important drawbacks: (i) the label ordering is decided at random, although it usually has a strong effect on predictive accuracy; (ii) all labels are inserted into the chain, although some of them might carry irrelevant information to discriminate the others. In this paper we tackle both problems at once, by proposing a novel genetic algorithm capable of searching for a single optimized label ordering, while at the same time taking into consideration the utilization of partial chains. Experiments on benchmark datasets demonstrate that our approach is able to produce models that are both simpler and more accurate.

Automatic classification of carbonate rocks permeability from 1H NMR relaxation data

This work investigates the effectiveness of data mining analysis on NMR data.The goal is to accurately predict the permeability class of carbonate rocks.Our approach outperforms the traditional NMR models Timur-Coates and Kenyon.Traditional models ignore the singular relationship between T2 bins and pore throat.Data mining models capture the influence of each T2 bin over the permeability class. The accurate permeability mapping, even with the aid of modern borehole geophysics methods, is still a big challenge on the reservoir management framework. One concern within the petrophysics community is that rock permeability value predicted by well logging should not be considered as absolute, mainly for carbonates, but a relative index for identifying more permeable zones. Therefore, in this paper a permeability classification methodology, based exclusively on 1H NMR (Nuclear Magnetic Resonance) relaxation data, was evaluated for the first time as an alternative to the prediction of permeability as a continuous variable. To pursue this, a side-by-side comparison of different data mining techniques for the permeability classification task was performed using a petrophysical dataset with 78 rock samples from six different carbonate reservoirs. The effectiveness of six classification algorithms (k-NN, Naive Bayes, C4.5, SMO, Random Forest and Multilayer Perceptron) was evaluated to predict the rock permeability class according to the following ranges: low ( 100mD). Discretization and feature selection strategies were also employed as preprocessing steps in order to improve the classification accuracy. For the studied dataset, the results demonstrated that the Random Forest and SMO strategies delivered the best classification performance among the selected classifiers. The computational experiments also evidenced that our approach led to more accurate predictions when compared with two methods widely adopted by the petroleum industry (Kenyon and Timur-Coates models).

Mining exceptions in databases

This paper addresses the problem of mining exceptions from multidimensional databases The goal of our proposed model is to find association rules that become weaker in some specific subsets of the database The candidates for exceptions are generated combining previously discovered multidimensional association rules with a set of significant attributes specified by the user The exceptions are mined only if the candidates do not achieve an expected support We describe a method to estimate these expectations and propose an algorithm that finds exceptions Experimental results are also presented.

Mineração de Exceções Aplicada aos Sistemas para Detecção de Intrusões

Os sistemas para a deteccao de intrusoes em redes de computadores frequentemente utilizam modelos baseados em regras para o reconhecimento de padroes suspeitos nos dados do trafego. Este trabalho apresenta uma tecnica baseada na mineracao de excecoes que pode ser utilizada para aumentar a eficiencia deste tipo de sistema. As excecoes representam regras de associacao que tornam-se extremamente fortes (excecoes positivas) ou extremamente fracas (excecoes negativas) em subconjuntos de uma base de dados que satisfazem condicoes especificas sobre atributos selecionados. Sao apresentados os resultados obtidos a partir da aplicacao desta tecnica sobre a base KDDCup99 que registra informacoes sobre conexoes de rede.