Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ LAReferencia - Red F...arrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
addClaim

This Research product is the result of merged Research products in OpenAIRE.

You have already added 0 works in your ORCID record related to the merged Research product.

Uma contribuição ao problema de seleção de modelos neurais usando o princípio de máxima correlação dos erros

Authors: Medeiros, Cláudio Marques de Sá;

Uma contribuição ao problema de seleção de modelos neurais usando o princípio de máxima correlação dos erros

Abstract

This thesis proposes a new pruning method which eliminates redundant weights in a multilayer perceptron (MLP). Conventional pruning techniques, like Optimal Brain Surgeon (OBS) and Optimal Brain Damage (OBD), are based on weight sensitivity analysis, which requires the inversion of the error Hessian matrix of the loss function (i.e. mean squared error). This inversion is specially susceptible to numerical problems due to poor conditioning of the Hessian matrix and demands great computational efforts. Another kind of pruning method is based on the regularization of the loss function, but it requires the determination of the regularization parameter by trial and error. The proposed method is based on "Maximum Correlation Errors Principle" (MAXCORE). The idea in this principle is to evaluate the importance of each network connection by calculating the cross correlation among errors in a layer and the back-propagated errors in the preceding layer, starting from the output layer and working through the network until the input layer is reached. The connections which have larger correlations remain and the others are pruned from the network. The evident advantage of this procedure is its simplicity, since matrix inversion or parameter adjustment are not necessary. The performance of the proposed method is evaluated in pattern classifi cation tasks and the results are compared to those achieved by the OBS/OBD techniques and also by regularization-based method. For this purpose, arti ficial data sets are used to highlight some important characteristics of the proposed methodology. Furthermore, well known benchmarking data sets, such as IRIS, WINE and DERMATOLOGY, are also used for the sake of evaluation. A real-world biomedical data set related to pathologies of the vertebral column is also used. The results obtained show that the proposed method achieves equivalent or superior performance compared to conventional pruning methods, with the additional advantages of low computational cost and simplicity. The proposed method also presents e ficient behavior in pruning the input units, which suggests its use as a feature selection method.

Propõe-se nesta tese um método de poda de pesos para redes Perceptron Multicamadas (MLP). Técnicas clássicas de poda convencionais, tais como Optimal Brain Surgeon(OBS) e Optimal Brain Damage(OBD), baseiam-se na análise de sensibilidade de cada peso da rede, o que requer a determinação da inversa da matriz Hessiana da função-custo. A inversão da matriz Hessiana, além de possuir um alto custo computacional, é bastante susceptível a problemas numéricos decorrentes do mal-condicionamento da mesma. Métodos de poda baseados na regularização da função-custo, por outro lado, exigem a determinação por tentativa-e-erro de um parâmetro de regularização. Tendo em mente as limitações dos métodos de poda supracitados, o método proposto baseia-se no "Princípio da Máxima Correlação dos Erros" (MAXCORE). A idéia consiste em analisar a importância de cada conexão da rede a partir da correlação cruzada entre os erros em uma camada e os erros retropropagados para a camada anterior, partindo da camada de saída em direção à camada de entrada. As conexões que produzem as maiores correlações tendem a se manter na rede podada. Uma vantagem imediata deste procedimento está em não requerer a inversão de matrizes, nem um parâmetro de regularização. O desempenho do método proposto é avaliado em problemas de classi ficação de padrões e os resultados são comparados aos obtidos pelos métodos OBS/OBD e por um método de poda baseado em regularização. Para este fi m, são usados, além de dados arti cialmente criados para salientar características importantes do método, os conjuntos de dados bem conhecidos da comunidade de aprendizado de máquinas: Iris, Wine e Dermatology. Utilizou-se também um conjunto de dados reais referentes ao diagnóstico de patologias da coluna vertebral. Os resultados obtidos mostram que o método proposto apresenta desempenho equivalente ou superior aos métodos de poda convencionais, com as vantagens adicionais do baixo custo computacional e simplicidade. O método proposto também mostrou-se bastante agressivo na poda de unidades de entrada (atributos), o que sugere a sua aplicação em seleção de características.

Keywords

Teoria dos erros, Redes neurais (Computação), Teleinformática

  • BIP!
    Impact byBIP!
    citations
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
citations
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average