A survey of evolutionary algorithms for decision-tree induction

This paper presents a survey of evolutionary algorithms that are designed for decision-tree induction. In this context, most of the paper focuses on approaches that evolve decision trees as an alternate heuristics to the traditional top-down divide-and-conquer approach. Additionally, we present some alternative methods that make use of evolutionary algorithms to improve particular components of decision-tree classifiers. The paper's original contributions are the following. First, it provides an up-to-date overview that is fully focused on evolutionary algorithms and decision trees and does not concentrate on any specific evolutionary approach. Second, it provides a taxonomy, which addresses works that evolve decision trees and works that design decision-tree components by the use of evolutionary algorithms. Finally, a number of references are provided that describe applications of evolutionary algorithms for decision-tree induction in different domains. At the end of this paper, we address some important issues and open questions that can be the subject of future research.

Combining meta-learning and search techniques to select parameters for support vector machines

Support Vector Machines (SVMs) have achieved very good performance on different learning problems. However, the success of SVMs depends on the adequate choice of the values of a number of parameters (e.g., the kernel and regularization parameters). In the current work, we propose the combination of meta-learning and search algorithms to deal with the problem of SVM parameter selection. In this combination, given a new problem to be solved, meta-learning is employed to recommend SVM parameter values based on parameter configurations that have been successfully adopted in previous similar problems. The parameter values returned by meta-learning are then used as initial search points by a search technique, which will further explore the parameter space. In this proposal, we envisioned that the initial solutions provided by meta-learning are located in good regions of the search space (i.e. they are closer to optimum solutions). Hence, the search algorithm would need to evaluate a lower number of candidate solutions when looking for an adequate solution. In this work, we investigate the combination of meta-learning with two search algorithms: Particle Swarm Optimization and Tabu Search. The implemented hybrid algorithms were used to select the values of two SVM parameters in the regression domain. These combinations were compared with the use of the search algorithms without meta-learning. The experimental results on a set of 40 regression problems showed that, on average, the proposed hybrid methods obtained lower error rates when compared to their components applied in isolation.

Recent trends in intelligent data analysis

European Regional Development Fund

Evolving decision trees with beam search-based initialization and lexicographic multi-objective evaluation

Decision tree induction algorithms represent one of the most popular techniques for dealing with classification problems. However, traditional decision-tree induction algorithms implement a greedy approach for node splitting that is inherently susceptible to local optima convergence. Evolutionary algorithms can avoid the problems associated with a greedy search and have been successfully employed to the induction of decision trees. Previously, we proposed a lexicographic multi-objective genetic algorithm for decision-tree induction, named LEGAL-Tree. In this work, we propose extending this approach substantially, particularly w.r.t. two important evolutionary aspects: the initialization of the population and the fitness function. We carry out a comprehensive set of experiments to validate our extended algorithm. The experimental results suggest that it is able to outperform both traditional algorithms for decision-tree induction and another evolutionary algorithm in a variety of application domains.

Hierarchical multi-label classification using local neural networks

Hierarchical multi-label classification is a complex classification task where the classes involved in the problem are hierarchically structured and each example may simultaneously belong to more than one class in each hierarchical level. In this paper, we extend our previous works, where we investigated a new local-based classification method that incrementally trains a multi-layer perceptron for each level of the classification hierarchy. Predictions made by a neural network in a given level are used as inputs to the neural network responsible for the prediction in the next level. We compare the proposed method with one state-of-the-art decision-tree induction method and two decision-tree induction methods, using several hierarchical multi-label classification datasets. We perform a thorough experimental analysis, showing that our method obtains competitive results to a robust global method regarding both precision and recall evaluation measures.

BENTHIC MACROFAUNA STRUCTURE IN THE NORTHEAST AREA OF TODOS OS SANTOS BAY, BAHIA STATE, BRAZIL: PATTERNS OF SPATIAL AND SEASONAL DISTRIBUTION

Structure of intertidal and subtidal benthic macrofauna in the northeastern region of Todos os Santos Bay (TSB), northeast Brazil, was investigated during a period of two years. Relationships with environmental parameters were studied through uni- and multivariate statistical analyses, and the main distributional patterns shown to be especially related to sediment type and content of organic fractions (Carbon, Nitrogen, Phosphorus), on both temporal and spatial scales. Polychaete annelids accounted for more than 70% of the total fauna and showed low densities, species richness and diversity, except for the area situated on the reef banks. These banks constitute a peculiar environment in relation to the rest of the region by having coarse sediments poor in organic matter and rich in biodetritic carbonates besides an abundant and diverse fauna. The intertidal region and the shallower area nearer to the oil refinery RLAM, with sediments composed mainly of fine sand, seem to constitute an unstable system with few highly dominant species, such as Armandia polyophthalma and Laeonereis acuta. In the other regions of TSB, where muddy bottoms predominated, densities and diversity were low, especially in the stations near the refinery. Here the lowest values of the biological indicators occurred together with the highest organic compound content. In addition, the nearest sites (stations 4 and 7) were sometimes azoic. The adjacent Caboto, considered as a control area at first, presented low density but intermediate values of species diversity, which indicates a less disturbed environment in relation to the pelitic infralittoral in front of the refinery. The results of the ordination analyses evidenced five homogeneous groups of stations (intertidal; reef banks; pelitic infralittoral; mixed sediments; Caboto) with different specific patterns, a fact which seems to be mainly related to granulometry and chemical sediment characteristics.

CAPTURE FISHERY IN NORTHERN TODOS OS SANTOS BAY, TROPICAL SOUTHWESTERN ATLANTIC, BRAZIL

Information on marine and estuarine capture fishery activity in northern Todos os Santos Bay, northeastern Brazil, based on daily data collected between September 2003 and June 2005 is presented. Small-scale artisanal fishery in this area includes the use of traditional vessels both non-motorized and motorized for locomotion, being carried out mainly by canoe or on foot, and involves many different kinds of gear, including gillnet, hook and line, seine nets, and traps. A total of 113 taxa were grouped into 77 resources, including 88 fish, 10 crustaceans, and 15 mollusks. Data on nominal catches of fish, crustaceans and mollusks are presented by month and location. A total of 345.2 tonnes of fishery resources were produced (285.4 tonnes of fish, 39.2 tonnes of fresh invertebrates, and 20.6 tonnes of processed invertebrates). Temporal variation in the fish catch was associated with the life cycle of the species or with the hydrographic conditions. The first-sale value of this catch amounted to around US$ 615,000.00, fishes representing 71.3% of it. A table of the average price of each fishery resource is presented. The results produced in this study may be considered a reference for future monitoring programs of fishery resources in the area.

ICHTHYOFAUNA IN AN ESTUARY OF THE MATARIPE AREA, TODOS OS SANTOS BAY, BAHIA, BRAZIL

The community structure and dynamics as well as some biological parameters of selected species of the ichthyofauna of the Mataripe estuarine region affected by the Landulfo Alves Oil Refinery (RLAM) were analyzed. Twenty stations were sampled with a gillnet in five different periods: August and December 2003, March and July/August 2004 and January 2005. Thirty-five actinopterygian species and one elasmobranch species were recorded, Oligoplites saurus, Diapterus rhombeus, Lutjanus synagris and Scomberomorus brasiliensis among them, on all the campaigns. A total of 1368 specimens, weighing 36.10 kg, were caught. The ichthyofauna total biomass was greater, in weight, on the eastern side of the study region, especially at the stations close to the shoals/reefs and the rocky bottom. A similar pattern was also observed for the diversity values. In general, low evenness and diversity were observed in the area studied, possibly as a result of the fishing gear used. D. rhombeus juveniles dominated in all but one of the samplings (July 2004), in which latter Cyclichthys spinosus was dominant. Carangids and species associated with consolidated bottoms were observed, although in small numbers, throughout the study period. In spite of the limitations imposed by the gear used for sampling, the estuarine area influenced by the RLAM was seen to play a role as a growth area for the great majority of species, especially the mojarra (D. rhombeus), but it offers no fishing potential due to the prevalence of young and small individuals. Evidence of imminent spawning was recorded for Pomadasys corvinaeformis in August 2003, and recent spawning in March 2004 for Oligoplites saurus. Further, mature individuals occurred in insufficient numbers to permit population level evaluation.

Admiralty Bay Benthos Diversity-A census of a complex polar ecosystem

A thorough census of Admiralty Bay benthic biodiversity was completed through the synthesis of data, acquired from more than 30 years of observations. Most of the available records arise from successive Polish and Brazilian Antarctic expeditions organized since 1977 and 1982, respectively, but also include new data from joint collecting efforts during the International Polar Year (2007-2009). Geological and hydrological characteristics of Admiralty Bay and a comprehensive species checklist with detailed data on the distribution and nature of the benthic communities are provided. Approximately 1300 species of benthic organisms (excluding bacteria, fungi and parasites) were recorded from the bay`s entire depth range (0-500 m). Generalized classifications and the descriptions of soft-bottom and hard-bottom invertebrate communities are presented. A time-series analysis showed seasonal and interannual changes in the shallow benthic communities, likely to be related to ice formation and ice melt within the bay. As one of the best studied regions in the maritime Antarctic Admiralty Bay represents a legacy site, where continued, systematically integrated data sampling can evaluate the effects of climate change on marine life. Both high species richness and high assemblage diversity of the Admiralty Bay shelf benthic community have been documented against the background of habitat heterogeneity. (C) 2010 Elsevier Ltd. All rights reserved.

Bacterial diversity in rhizosphere soil from Antarctic vascular plants of Admiralty Bay, maritime Antarctica

The Antarctic is a pristine environment that contributes to the maintenance of the global climate equilibrium. The harsh conditions of this habitat are fundamental to selecting those organisms able to survive in such an extreme habitat and able to support the relatively simple ecosystems. The DNA of the microbial community associated with the rhizospheres of Deschampsia antarctica Desv (Poaceae) and Colobanthus quitensis (Kunth) BartI (Caryophyllaceae), the only two native vascular plants that are found in Antarctic ecosystems, was evaluated using a 16S rRNA multiplex 454 pyrosequencing approach. This analysis revealed similar patterns of bacterial diversity between the two plant species from different locations, arguing against the hypothesis that there would be differences between the rhizosphere communities of different plants. Furthermore, the phylum distribution presented a peculiar pattern, with a bacterial community structure different from those reported of many other soils. Firmicutes was the most abundant phylum in almost all the analyzed samples, and there were high levels of anaerobic representatives. Also, some phyla that are dominant in most temperate and tropical soils, such as Acidobacteria, were rarely found in the analyzed samples. Analyzing all the sample libraries together, the predominant genera found were Bifidobacterium (phylum Actinobacteria), Arcobacter (phylum Proteobacteria) and Faecalibacterium (phylum Firmicutes). To the best of our knowledge, this is the first major bacterial sequencing effort of this kind of soil, and it revealed more than expected diversity within these rhizospheres of both maritime Antarctica vascular plants in Admiralty Bay, King George Island, which is part of the South Shetlands archipelago. The ISME Journal (2010) 4, 989-1001; doi:10.1038/ismej.2010.35; published online 1 April 2010

SPATIAL AND TEMPORAL DISTRIBUTION OF FISH IN PALMAS BAY, UBATUBA, BRAZIL

The aim of this study was to analyze the distribution and abundance of the fish fauna of Palmas bay on Anchieta Island in southeastern Brazil. Specimens were caught in the summer and winter of 1992, using an otter trawl at three locations in the bay. The specimens were caught in both the nighttime and daytime. Data on the water temperature and salinity were recorded for the characterization of the predominant water mass in the region, and sediment samples were taken for granulometric analysis. A total of 7 656 specimens (79 species), with a total weight of approximately 300 kg, were recorded. The most abundant species were Eucinostomus argenteus, Ctenosciaena gracilicirrhus, Haemulon steindachneri, Eucinostomus gula and Diapterus rhombeus, which together accounted for more than 73% of the sample. In general, the ecological indices showed no differences in the composition of species for the abiotic variables analyzed. The multivariate analysis showed that the variations in the distribution of the fish fauna were mainly associated with intra-annual differences in temperature and salinity, resulting from the presence of South Atlantic Central Water (SACW) in the area during the summer. The analysis also showed an association with the type of bottom and a lesser association with respect to the night/day periods.

Occurrence of organochlorine compounds in Euphausia superba and unhatched eggs of Pygoscelis genus penguins from Admiralty Bay (King George Island, Antarctica) and estimation of biomagnification factors

Polychlorinated biphenyls (PCBs) and organochlorine pesticides are compounds that do not occur naturally in the environment and are not easily degraded by chemical or microbiological action. In the present work, those compounds were analysed in unhatched penguin eggs and whole krill collected in Admiralty Bay, King George Island, Antarctica in the austral summers of 2004-2005 and 2005-2006. The compounds found in higher levels (in a wet weight basis) were, in most of the egg samples, the PCBs (2.53-78.7 ng g(-1)), DDTs (2.07-38.0 ng g(-1)) and HCB (4.99-39.1 ng g(-1)) and after Kruskal-Wallis ANOVA, the occurrence seemed to be species-specific for the Pygoscelis genus. In all of the cases, the levels found were not higher than the ones in Arctic birds in a similar trophic level. The krill samples analysis allowed estimating the biomagnification factors (which resulted in up to 363 for HCB, one order of magnitude higher than DDTs and chlordanes and two orders of magnitude higher than the other groups) of the compounds found in eggs, whose only source of contamination is the female-offspring transfer. (C) 2009 Elsevier Ltd. All rights reserved.

Morphodynamics of structurally controlled headland-bay beaches in southeastern Brazil: A review

This paper presents a comprehensive review on the interaction between hydrodynamic processes, beach morphology and sedimentology at large scale coastal behaviour along the coastline of Santa Catarina, between Laguna and Sao Francisco Island, a microtidal east coast swell environment with headland and bay geomorphologies. The parabolic bay shape equation has proven to be a convenient and practical tool for studying the stability of the headland-bay beaches, tombolos, and salients in Santa Catarina. The beaches exhibit different patterns of sediment removal as a function of the degree of beach curvature. In highly curved beaches, there is a well-developed shadow zone and a range of morphodynamic conditions, from a sheltered low-energy beach adjacent to the downdrift headland to a high-energy exposed beach on the straight end of the headland-bay beach. The less curved beaches instead, tend to show more uniform behaviour since they are directly exposed to incident waves. There is no obvious relationship between average wave height and mean grain size, showing the importance of sediment source to characterize the sedimentary distribution patterns in the study area. The analysis of beaches showed that beach morphodynamics and sequence profiles for a bay-headland coast in a microtidal east coast environment is a function of geological inheritance (e.g., distance between headlands and orientation, nearshore and inner shelf morphology, coastal plain morphology, and sediment source), and hydrodynamic factors (wave conditions, oceanic wave exposure and relative tidal range). (C) 2009 Elsevier B.V. All rights reserved.

A snapshot of suspended sediment and fluid mud occurrence in a mixed-energy embayment, Tijucas Bay, Brazil

Along the southern Brazilian coast, Tijucas Bay is known for its unique muddy tidal flats associated with chenier plains. Previous field observations pointed to very high suspended sediment concentrations (SSCs) in the inner parts of the bay, and in the estuary of the Tijucas River, suggesting the presence of fluid mud. In this study, the occurrences of suspended sediments and fluid mud were examined during a larger-scale, high-resolution 2-day field campaign on 1-2 May 2007, encompassing survey lines spanning nearly 80 km, 75 water sampling stations for near-bottom density estimates, and ten sediment sampling stations. Wave refraction modeling provided qualitative wave energy estimates as a function of different incidence directions. The results show that SSC increases toward the inner bay near the water surface, but seaward near the bottom. This suggests that suspended sediment is supplied by the local rivers, in particular the Tijucas. Near-surface SSCs were of the order of 50 mg l(-1) close to the shore, but exceeded 100 mg l(-1) near the bottom in the deeper parts of the bay. Fluid mud thickness and location given by densimetry and echo-sounding agreed in some places, although being mostly discordant. The best agreement was observed where wave energy was high during the campaign. The discrepancy between the two methods may be an indication for the existence of fluid mud, which is recorded by one method but not the other. Agreement is considered to be an indication of fluidization, whereas disagreement indicates more consolidation. Wave modeling suggests that waves from the ENE and SE are the most effective in supplying energy to the inner bay, which may induce the liquefaction of mud deposits to form fluid mud. Nearshore mud resuspension and weak horizontal currents result in sediment-laden offshore flow, which explains the higher SSCs measured in the deeper parts of the bay, besides providing a mechanism for fine-sediment export to the inner shelf.

Historical record of polycyclic aromatic hydrocarbons (PAHs) and spheroidal carbonaceous particles (SCPs) in marine sediment cores from Admiralty Bay, King George Island, Antarctica

This paper describes the first results of polycyclic aromatic hydrocarbons (PAHs) and spheroidal carbonaceous particles (SCPs) in sediment cores of Admiralty Bay, Antarctica. These markers were used to assess the local input of anthropogenic materials (particulate and organic compounds) as a result of the influence of human occupation in a sub-Antarctic region and a possible long-range atmospheric transport of combustion products from sources in South America. The highest SCPs and PAHs concentrations were observed during the last 30 years, when three research stations were built in the area and industrial activities in South America increased. The concentrations of SCPs and PAHs were much lower than those of other regions in the northern hemisphere and other reported data for the southern hemisphere. The PAH isomer ratios showed that the major sources of PAHs are fossil fuels/petroleum, biomass combustion and sewage contribution generally close to the Brazilian scientific station. (C) 2009 Elsevier Ltd. All rights reserved.