We study the issue of error diversity in ensembles of neural networks. In ensembles of
regression estimators, the measurement of diversity can be formalised as the Bias-VarianceCovariance
decomposition. In ensembles of classifiers, there is no neat theory in the literature
to date. Our objective is to understand how to precisely define, measure, and
create diverse errors for both cases. As a focal point we study one algorithm, Negative Correlation
(NC) Learning which claimed, and showed empirical evidence, to enforce useful
error diversity, creating neural network ensembles with very competitive performance on
both classification and regression problems. With the lack of a solid understanding of its
dynamics, we engage in a theoretical and empirical investigation.
In an initial empirical stage, we demonstrate the application of an evolutionary search
algorithm to locate the optimal value for λ, the configurable parameter in NC. We observe
the behaviour of the optimal parameter under different ensemble architectures and datasets;
we note a high degree of unpredictability, and embark on a more formal investigation.
During the theoretical investigations, we find that NC succeeds due to exploiting the
Ambiguity decomposition of mean squared error. We provide a grounding for NC in a
statistics context of bias, variance and covariance, including a link to a number of other
algorithms that have exploited Ambiguity. The discoveries we make regarding NC are not
limited to neural networks. The majority of observations we make are in fact properties of
the mean squared error function. We find that NC is therefore best viewed as a framework,
rather than an algorithm itself, meaning several other learning techniques could make use
of it.
We study the issue of error diversity in ensembles of neural networks. In ensembles ofregression estimators, the measurement of diversity can be formalised as the Bias-VarianceCovariancedecomposition. In ensembles of classifiers, there is no neat theory in the literatureto date. Our objective is to understand how to precisely define, measure, andcreate diverse errors for both cases. As a focal point we study one algorithm, Negative Correlation(NC) Learning which claimed, and showed empirical evidence, to enforce usefulerror diversity, creating neural network ensembles with very competitive performance onboth classification and regression problems. With the lack of a solid understanding of itsdynamics, we engage in a theoretical and empirical investigation.In an initial empirical stage, we demonstrate the application of an evolutionary searchalgorithm to locate the optimal value for λ, the configurable parameter in NC. We observethe behaviour of the optimal parameter under different ensemble architectures and datasets;we note a high degree of unpredictability, and embark on a more formal investigation.During the theoretical investigations, we find that NC succeeds due to exploiting theAmbiguity decomposition of mean squared error. We provide a grounding for NC in astatistics context of bias, variance and covariance, including a link to a number of otheralgorithms that have exploited Ambiguity. The discoveries we make regarding NC are notlimited to neural networks. The majority of observations we make are in fact properties ofthe mean squared error function. We find that NC is therefore best viewed as a framework,rather than an algorithm itself, meaning several other learning techniques could make useof it.
การแปล กรุณารอสักครู่..
