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.

Two series of new semisynthetic triterpene derivatives: differences in anti-malarial activity, cytotoxicity and mechanism of action

Background The discovery and development of anti-malarial compounds of plant origin and semisynthetic derivatives thereof, such as quinine (QN) and chloroquine (CQ), has highlighted the importance of these compounds in the treatment of malaria. Ursolic acid analogues bearing an acetyl group at C-3 have demonstrated significant anti-malarial activity. With this in mind, two new series of betulinic acid (BA) and ursolic acid (UA) derivatives with ester groups at C-3 were synthesized in an attempt to improve anti-malarial activity, reduce cytotoxicity, and search for new targets. In vitro activity against CQ-sensitive Plasmodium falciparum 3D7 and an evaluation of cytotoxicity in a mammalian cell line (HEK293T) are reported. Furthermore, two possible mechanisms of action of anti-malarial compounds have been evaluated: effects on mitochondrial membrane potential (ΔΨm) and inhibition of β-haematin formation. Results Among the 18 derivatives synthesized, those having shorter side chains were most effective against CQ-sensitive P. falciparum 3D7, and were non-cytotoxic. These derivatives were three to five times more active than BA and UA. A DiOC6(3) ΔΨm assay showed that mitochondria are not involved in their mechanism of action. Inhibition of β-haematin formation by the active derivatives was weaker than with CQ. Compounds of the BA series were generally more active against P. falciparum 3D7 than those of the UA series. Conclusions Three new anti-malarial prototypes were obtained from natural sources through an easy and relatively inexpensive synthesis. They represent an alternative for new lead compounds for anti-malarial chemotherapy.

Study of cerebral gene expression densities using Voronoi analysis

As the available public cerebral gene expression image data increasingly grows, the demand for automated methods to analyze such large amount of data also increases. An important study that can be carried out on these data is related to the spatial relationship between gene expressions. Similar spatial density distribution of expression between genes may indicate they are functionally correlated, thus the identification of these similarities is useful in suggesting directions of investigation to discover gene interactions and their correlated functions. In this paper, we describe the use of a high-throughput methodology based on Voronoi diagrams to automatically analyze and search for possible local spatial density relationships between gene expression images. We tested this method using mouse brain section images from the Allen Mouse Brain Atlas public database. This methodology provided measurements able to characterize the similarity of the density distribution between gene expressions and allowed the visualization of the results through networks and Principal Component Analysis (PCA). These visualizations are useful to analyze the similarity level between gene expression patterns, as well as to compare connection patterns between region networks. Some genes were found to have the same type of function and to be near each other in the PCA visualizations. These results suggest cerebral density correlations between gene expressions that could be further explored. (C) 2011 Elsevier B.V. All rights reserved.

Antidepressants prevent hierarchy destabilization induced by lipopolysaccharide administration in mice: a neurobiological approach to depression

In spite of the high prevalence and negative impact of depression, little is known about its pathophysiology. Basic research on depression needs new animal models in order to increase knowledge of the disease and search for new therapies. The work presented here aims to provide a neurobiologically validated model for investigating the relationships among sickness behavior, antidepressants treatment, and social dominance behavior. For this purpose, dominant individuals from dyads of male Swiss mice were treated with the bacterial endotoxin lipopolysaccharide (LPS) to induce social hierarchy destabilization. Two groups were treated with the antidepressants imipramine and fluoxetine prior to LPS administration. In these groups, antidepressant treatment prevented the occurrence of social destabilization. These results indicate that this model could be useful in providing new insights into the understanding of the brain systems involved in depression.

On the search for potential anti-Trypanosoma cruzi drugs: Synthesis and biological evaluation of 2-hydroxy-3-methylamino and 1,2,3-triazolic naphthoquinoidal compounds obtained by click chemistry reactions

Five 2-hydroxy-3-substituted-aminomethyl naphthoquinones, nine 1,2,3-triazolic para-naphthoquinones, five nor-beta-lapachone-based 1,2,3-triazoles, and several other naphthoquinonoid compounds were synthesized and evaluated against the infective bloodstream form of Trypanosoma cruzi, the etiological agent of Chagas disease, continuing our screening program for new trypanocidal compounds. Among all the substances, 16-18, 23, 25-29 and 30-33 were herein described for the first time and fifteen substances were identified as more potent than the standard drug benznidazole, with IC50/24 h values in the range of 10.9-101.5 mu M. Compounds 14 and 19 with Selectivity Index of 18.9 and 6.1 are important structures for further studies. (C) 2012 Elsevier Masson SAS. All rights reserved.

SEARCH FOR POINT-LIKE SOURCES OF ULTRA-HIGH ENERGY NEUTRINOS AT THE PIERRE AUGER OBSERVATORY AND IMPROVED LIMIT ON THE DIFFUSE FLUX OF TAU NEUTRINOS

The surface detector array of the Pierre Auger Observatory can detect neutrinos with energy E-nu between 10(17) eV and 10(20) eV from point-like sources across the sky south of +55 degrees and north of -65 degrees declinations. A search has been performed for highly inclined extensive air showers produced by the interaction of neutrinos of all flavors in the atmosphere (downward-going neutrinos), and by the decay of tau leptons originating from tau neutrino interactions in Earth's crust (Earth-skimming neutrinos). No candidate neutrinos have been found in data up to 2010 May 31. This corresponds to an equivalent exposure of similar to 3.5 years of a full surface detector array for the Earth-skimming channel and similar to 2 years for the downward-going channel. An improved upper limit on the diffuse flux of tau neutrinos has been derived. Upper limits on the neutrino flux from point-like sources have been derived as a function of the source declination. Assuming a differential neutrino flux k(PS) . E-nu(-2). from a point-like source, 90% confidence level upper limits for k(PS) at the level of approximate to 5x10(-7) and 2.5x10(-6) GeV cm(-2) s(-1) have been obtained over a broad range of declinations from the searches for Earth-skimming and downward-going neutrinos, respectively.

Search for Lorentz invariance and CPT violation with muon antineutrinos in the MINOS Near Detector

We have searched for sidereal variations in the rate of antineutrino interactions in the MINOS Near Detector. Using antineutrinos produced by the NuMI beam, we find no statistically significant sidereal modulation in the rate. When this result is placed in the context of the Standard Model Extension theory we are able to place upper limits on the coefficients defining the theory. These limits are used in combination with the results from an earlier analysis of MINOS neutrino data to further constrain the coefficients.

CARBON MARKET AND INSTITUTIONS: OPPORTUNITIES IN THE SEARCH OF A NEW MODEL FOR DEVELOPMENT

Seeking alternatives for the economic system to face the several crises it has gone through lately (electrical power, cultural, financing and technological) brought about a new market involving the Kyoto Protocol signatory countries: the carbon market. The present article aims at assessing the carbon market institutional issue in Brazil by identifying the risks and opportunities inherent to the institutional agent characteristics and to that market rules. The research methodology was bibliographic and based on the analysis of the Securities and Exchange Commission of Brazil (Comissao de Valores Mobiliarios and Bolsa Mercantil de Valores) contents. Its theoretical basis rests on concepts of the institution and the new institutional economy. The results show that in spite of the risks and institutional problems it involves, the carbon market is promising due to the opportunities create by new technologies and energies developed to achieve and sustain the capitalist system new cycle, addressed to produce a clean development.

A theoretical study of SnF2+, SnCl2+, and SnO2+ and their experimental search

We present a detailed theoretical study of the stability of the gas-phase diatomic dications SnF2+, SnCl2+, and SnO2+ using ab initio computer calculations. The ground states of SnF2+, SnCl2+, and SnO2+ are thermodynamically stable, respectively, with dissociation energies of 0.45, 0.30, and 0.42 eV. Whereas SnF2+ dissociates into Sn2+ + F, the long range behaviour of the potential energy curves of SnCl2+ and SnO2+ is repulsive and wide barrier heights due to avoided crossing act as a kind of effective dissociation energy. Their equilibrium internuclear distances are 4.855, 5.201, and 4.852 a(0), respectively. The double ionisation energies (T-e) to form SnF2+, SnCl2+, and SnO2+ from their respective neutral parents are 25.87, 23.71, and 25.97 eV. We combine our theoretical work with the experimental results of a search for these doubly positively charged diatomic molecules in the gas phase. SnO2+ and SnF2+ have been observed for prolonged oxygen (O-16(-)) ion beam sputtering of a tin metal foil and of tin (II) fluoride (SnF2) powder, respectively, for ion flight times of about 10(-5) s through a magnetic-sector mass spectrometer. In addition, SnCl2+ has been detected for O-16(-) ion surface bombardment of stannous (tin (II)) chloride (SnCl2) powder. To our knowledge, SnF2+ is a novel gas-phase molecule, whereas SnCl2+ had been detected previously by electron-impact ionization mass spectrometry, and SnO2+ had been observed before by spark source mass spectrometry as well as by atom probe mass spectrometry. We are not aware of any previous theoretical studies of these molecular systems. (C) 2012 American Institute of Physics. [http://dx.doi.org/10.1063/1.4758475]

Movement of Heterorhabditis amazonensis and Steinernema arenarium in search of corn fall armyworm larvae in artificial conditions

Spodoptera frugiperda (Smith, 1797) (Lepidoptera: Noctuidae) is considered to be the main pest of maize crops in Brazil. Entomopathogenic nematodes (EPN) may be used to control this pest and exhibit different, unique abilities to search for their hosts. The movement of EPN in relation to S. frugiperda was evaluated. To test for horizontal movement, a styrofoam enclosure filled with sand was divided into segments, nematodes were placed at the entrance to the enclosure and a larva was placed at the end of each division. The same approach was used to evaluate vertical movement; however, PVC pipes were used in this case. In general, the mortality was inversely proportional to the initial distance between host and nematodes. In the vertical displacement test, both nematodes were able to kill the larvae up to a distance of 25 cm. Therefore, the infective juveniles of H. amazonensis and S. arenarium can search out, infect and kill larvae of S. frugiperda at distances of up to 60 cm and 25 cm of horizontal and vertical displacement, respectively.

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.