Particle competition and cooperation to prevent error propagation from mislabeled data in semi-supervised learning

Semi-supervised learning is applied to classification problems where only a small portion of the data items is labeled. In these cases, the reliability of the labels is a crucial factor, because mislabeled items may propagate wrong labels to a large portion or even the entire data set. This paper aims to address this problem by presenting a graph-based (network-based) semi-supervised learning method, specifically designed to handle data sets with mislabeled samples. The method uses teams of walking particles, with competitive and cooperative behavior, for label propagation in the network constructed from the input data set. The proposed model is nature-inspired and it incorporates some features to make it robust to a considerable amount of mislabeled data items. Computer simulations show the performance of the method in the presence of different percentage of mislabeled data, in networks of different sizes and average node degree. Importantly, these simulations reveals the existence of the critical points of the mislabeled subset size, below which the network is free of wrong label contamination, but above which the mislabeled samples start to propagate their labels to the rest of the network. Moreover, numerical comparisons have been made among the proposed method and other representative graph-based semi-supervised learning methods using both artificial and real-world data sets. Interestingly, the proposed method has increasing better performance than the others as the percentage of mislabeled samples is getting larger. © 2012 IEEE.

Interação genótipo ambiente para perímetro escrotal em bovinos da raça Nelore utilizando normas de reação

Pós-graduação em Zootecnia - FCAV

Desenvolvimento de um sistema de gerenciamento de energia (EMS - Energy Management System) para a rede elétrica inteligente (Smart Grid)

Pós-graduação em Engenharia Elétrica - FEIS

Análise dos sistemas de produção dos estabelecimentos rurais do município de Palmital/SP em busca de estratégias para o desenvolvimento rural

Pós-graduação em Agronegócio e Desenvolvimento - Tupã

O discurso sobre a aula de matemática: articulando vozes na revista Nova Escola

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Simulations under Ideal and Nonideal Conditions for Characterization of a Passive Doppler Geographical Location System Using Extension of Data Reception Network

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Improving the passive doppler geolocation system through extension of the data reception network

O objetivo deste trabalho e apresentar uma investigação preliminar da precisão nos resultados do sistema de localização geográfica de transmissores desenvolvido utilizando o software da rede brasileira de coleta de dados. Um conjunto de medidas de desvio Doppler de uma única passagem do satélite, considerando uma Plataforma de Coleta de Dados (PCD) e uma rede de estações de recepção terrestrês, e denominado uma rede de recepção de dados. Assim, a rede brasileira de coleta de dados com o uso de múltiplas estações de recepção permitira o incremento na quantidade de dados coletados com consequente melhora na precisão e na confiabilidade das localizações fornecidas. Consequentemente uma maior quantidade de localizações válidas e mais precisas. Os resultados e análises foram obtidos sob duas condições: na primeira foi considerada uma condição prática com dados reais e dados ideais simulados, para comparar os resultados considerando a mesma passagem do satélite, transmissor e duas estações de recepção conhecidas; na segunda foram consideradas as condições ideais simuladas a partir de medidas de um transmissor fixo, três estações de recepção e dois satélites. Os resultados utilizando a rede de recepção de dados foram bastante satisfatórios. O estudo realizado mostrou a importãncia da instalação de novas estações de recepção terrenas distribuídas no territorio nacional, para um aumento na quantidade de medidas e consequentemente uma maior quantidade de localizações válidas e mais precisas.

GDP 4.0 transfer to SGP 3.0 for SCIAMACHY no2 column processing: Verification with SDOAS / GDOAS prototype algorithms and delta-validation with NDACC / UV-visible network data

Until mid 2006, SCIAMACHY data processors for the operational retrieval of nitrogen dioxide (NO2) column data were based on the historical version 2 of the GOME Data Processor (GDP). On top of known problems inherent to GDP 2, ground-based validations of SCIAMACHY NO2 data revealed issues specific to SCIAMACHY, like a large cloud-dependent offset occurring at Northern latitudes. In 2006, the GDOAS prototype algorithm of the improved GDP version 4 was transferred to the off-line SCIAMACHY Ground Processor (SGP) version 3.0. In parallel, the calibration of SCIAMACHY radiometric data was upgraded. Before operational switch-on of SGP 3.0 and public release of upgraded SCIAMACHY NO2 data, we have investigated the accuracy of the algorithm transfer: (a) by checking the consistency of SGP 3.0 with prototype algorithms; and (b) by comparing SGP 3.0 NO2 data with ground-based observations reported by the WMO/GAW NDACC network of UV-visible DOAS/SAOZ spectrometers. This delta-validation study concludes that SGP 3.0 is a significant improvement with respect to the previous processor IPF 5.04. For three particular SCIAMACHY states, the study reveals unexplained features in the slant columns and air mass factors, although the quantitative impact on SGP 3.0 vertical columns is not significant.

Preprocessing data for short-term load forecasting with a general regression neural network and a moving average filter

This paper proposes a filter based on a general regression neural network and a moving average filter, for preprocessing half-hourly load data for short-term multinodal load forecasting, discussed in another paper. Tests made with half-hourly load data from nine New Zealand electrical substations demonstrate that this filter is able to handle noise, missing data and abnormal data. © 2011 IEEE.

Semi-supervised learning with concept drift using particle dynamics applied to network intrusion detection data

Concept drift, which refers to non stationary learning problems over time, has increasing importance in machine learning and data mining. Many concept drift applications require fast response, which means an algorithm must always be (re)trained with the latest available data. But the process of data labeling is usually expensive and/or time consuming when compared to acquisition of unlabeled data, thus usually only a small fraction of the incoming data may be effectively labeled. Semi-supervised learning methods may help in this scenario, as they use both labeled and unlabeled data in the training process. However, most of them are based on assumptions that the data is static. Therefore, semi-supervised learning with concept drifts is still an open challenging task in machine learning. Recently, a particle competition and cooperation approach has been developed to realize graph-based semi-supervised learning from static data. We have extend that approach to handle data streams and concept drift. The result is a passive algorithm which uses a single classifier approach, naturally adapted to concept changes without any explicit drift detection mechanism. It has built-in mechanisms that provide a natural way of learning from new data, gradually "forgetting" older knowledge as older data items are no longer useful for the classification of newer data items. The proposed algorithm is applied to the KDD Cup 1999 Data of network intrusion, showing its effectiveness.

Comparison of logistic and neural network models to fit to the egg production curve of White Leghorn hens

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Strategies to promote farm animal welfare in Latin America and their effects on carcass and meat quality traits

Several initiatives, including research and development, increasing stakeholders' awareness and application of legislation and recommendation, have been carried out in Latin America to promote animal welfare and meat quality. Most activities focused on the impact of pre-slaughter conditions (facilities, equipment and handling procedures) on animal welfare and meat quality. The results are encouraging; data from Brazil, Chile and Uruguay showed that the application of the improved pre-slaughter handling practices reduced aggressive handling and the incidence of bruised carcasses at slaughter in cattle and pigs. These outcomes stimulated some to apply animal welfare concepts in livestock handling within the meat production chain as shown by the increasing demand for personnel training on the best. To attend this demand is important to expand local studies on farm animal welfare and to set up (or maintain) an efficient system for knowledge transfer to all stakeholders in the Latin America meat production chains. However, it is clear that to promote the long-term progress in this field is important to deliver practical solutions, assuring that they match the technical and financial conditions of those who are the target of training programs. (C) 2012 Elsevier Ltd. All rights reserved.

Influence of data structure on the estimation of the additive genetic direct and maternal covariance for early growth traits in Nellore cattle

The objective of the present study was to investigate the effect of data structure on estimated genetic parameters and predicted breeding values of direct and maternal genetic effects for weaning weight (WW) and weight gain from birth to weaning (BWG), including or not the genetic covariance between direct and maternal effects. Records of 97,490 Nellore animals born between 1993 and 2006, from the Jacarezinho cattle raising farm, were used. Two different data sets were analyzed: DI_all, which included all available progenies of dams without their own performance; DII_all, which included DI_all + 20% of recorded progenies with maternal phenotypes. Two subsets were obtained from each data set (DI_all and DII_all): DI_1 and DII_1, which included only dams with three or fewer progenies; DI_5 and DII_5, which included only dams with five or more progenies. (Co)variance components and heritabilities were estimated by Bayesian inference through Gibbs sampling using univariate animal models. In general, for the population and traits studied, the proportion of dams with known phenotypic information and the number of progenies per dam influenced direct and maternal heritabilities, as well as the contribution of maternal permanent environmental variance to phenotypic variance. Only small differences were observed in the genetic and environmental parameters when the genetic covariance between direct and maternal effects was set to zero in the data sets studied. Thus, the inclusion or not of the genetic covariance between direct and maternal effects had little effect on the ranking of animals according to their breeding values for WW and BWG. Accurate estimation of genetic correlations between direct and maternal genetic effects depends on the data structure. Thus, this covariance should be set to zero in Nellore data sets in which the proportion of dams with phenotypic information is low, the number of progenies per dam is small, and pedigree relationships are poorly known. (c) 2012 Elsevier B.V. All rights reserved.

Integration of a regional GPS network within ITRF using precise point positioning

GPS precise point positioning (PPP) can provide high precision 3-D coordinates. Combined pseudorange and carrier phase observables, precise ephemeris and satellite clock corrections, together with data from dual frequency receivers, are the key factors for providing such levels of precision (few centimeters). In general, results obtained from PPP are referenced to an arbitrary reference frame, realized from a previous free network adjustment, in which satellite state vectors, station coordinates and other biases are estimated together. In order to obtain consistent results, the coordinates have to be transformed to the relevant reference frame and the appropriate daily transformation parameters must be available. Furthermore, the coordinates have to be mapped to a chosen reference epoch. If a velocity field is not available, an appropriated model, such as NNR-NUVEL-IA, has to be used. The quality of the results provided by this approach was evaluated using data from the Brazilian Network for Continuous Monitoring of the Global Positioning System (RBMC), which was processed using GIPSY-OASIS 11 software. The results obtained were compared to SIRGAS 1995.4 and ITRF2000, and reached precision better than 2cm. A description of the fundamentals of the PPP approach and its application in the integration of regional GPS networks with ITRF is the main purpose of this paper.

A dense GNSS meteorological network for observing deep convection in the Amazon

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)