Results 1  10
of
12
When Networks Disagree: Ensemble Methods for Hybrid Neural Networks
, 1993
"... This paper presents a general theoretical framework for ensemble methods of constructing significantly improved regression estimates. Given a population of regression estimators, we construct a hybrid estimator which is as good or better in the MSE sense than any estimator in the population. We argu ..."
Abstract

Cited by 290 (2 self)
 Add to MetaCart
This paper presents a general theoretical framework for ensemble methods of constructing significantly improved regression estimates. Given a population of regression estimators, we construct a hybrid estimator which is as good or better in the MSE sense than any estimator in the population. We argue that the ensemble method presented has several properties: 1) It efficiently uses all the networks of a population  none of the networks need be discarded. 2) It efficiently uses all the available data for training without overfitting. 3) It inherently performs regularization by smoothing in functional space which helps to avoid overfitting. 4) It utilizes local minima to construct improved estimates whereas other neural network algorithms are hindered by local minima. 5) It is ideally suited for parallel computation. 6) It leads to a very useful and natural measure of the number of distinct estimators in a population. 7) The optimal parameters of the ensemble estimator are given in clo...
Improving Regression Estimation: Averaging Methods for Variance Reduction with Extensions to General Convex Measure Optimization
, 1993
"... ..."
Robust estimation of population size in closed animal populations from capture~recapture experiments
, 1983
"... This paper considers the problem of finding robust estimators of population size in closed Ksample capturerecapture experimerts.Particular attention is paid to models where heterogeneity of capture probabilities is allowed. First a general estimation procedure is given which does not depend on ass ..."
Abstract

Cited by 7 (0 self)
 Add to MetaCart
This paper considers the problem of finding robust estimators of population size in closed Ksample capturerecapture experimerts.Particular attention is paid to models where heterogeneity of capture probabilities is allowed. First a general estimation procedure is given which does not depend on assuming anything about the form of the distribution of capture probabilities. This is followed by a detailed discussion of the usefulness of the generalized jackknife technique to reduce bias. Numerical comparisons of the bias and variance of various estimators are given. Finally a general discussion is given with several recommendations on estimators to be used in practice. Key words: Capturerecapture sampling; Population size estimation; Heterogeneity;
IMPROVED RATIO TYPE ESTIMATOR USING JACK KNIFE METHOD OF ESTIMATION
"... In this paper, we propose to use an improved sampling strategy based on the modified ratio estimator using the population coefficient of variation and the coefficient of kurtosis of the auxiliary variable by Upadhyay and Singh (1999) for estimating the population mean (total) of the study variable i ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
In this paper, we propose to use an improved sampling strategy based on the modified ratio estimator using the population coefficient of variation and the coefficient of kurtosis of the auxiliary variable by Upadhyay and Singh (1999) for estimating the population mean (total) of the study variable in a finite population. Also the proposed sampling strategy is shown to be better in the sense of unbiased and smaller mean square error. A generalized JackKnife estimator is proposed and it is shown that the proposed JackKnife estimator is unbiased to the first order of approximation. A comparative study is made with usual sampling strategies utilizing the optimizing value of the characterizing scalar.
unknown title
"... Abstract Let f(x) be the density of a design variable X and m(x) = E[Y jX = x] the regression function. Then m(x) = G(x)=f(x), where G(x) = m(x)f(x). The Dirac ffifunctionis used to define a generalized empirical function Gn(x) for G(x) whose expectation equals G(x). This generalized empirical ..."
Abstract
 Add to MetaCart
Abstract Let f(x) be the density of a design variable X and m(x) = E[Y jX = x] the regression function. Then m(x) = G(x)=f(x), where G(x) = m(x)f(x). The Dirac ffifunctionis used to define a generalized empirical function Gn(x) for G(x) whose expectation equals G(x). This generalized empirical function exists only in the space of Schwartz distributions,so we introduce a local polynomial of order p approximation to Gn(\Delta) which provides estimators of the function G(x) and its derivatives. The density f(x) can be estimated in asimilar manner. The resulting local generalized empirical estimator (LGE) of m(x) is exactly the NadarayaWatson estimator at interior points when p = 1, but on the boundary theestimator automatically corrects the boundary effect. Asymptotic normality of the estimator is established. Asymptotic expressions for the mean squared errors are obtained and usedin bandwidth selection. Boundary behavior of the estimators is investigated in details. We use Monte Carlo simulations to show that the proposed estimator with p = 1 comparesfavorably with the NadarayaWatson and the popular local linear regression smoother. Keywords: boundary adaptive, Dirac ffifunction, local polynomial, local empirical, NadarayaWatson estimator. DOI: 10.1360/02ys0376 Given i.i.d. observations fXi; Yig n i=1 of (X; Y), consider the estimation of the regression function m(x) = E[Y jX = x]: Nadaraya [1] and Watson[2] proposed to estimate the regression function m(x) via the following kernel approach: ^m(x) =
Collusion Program' in The U.S. Crop Insurance Applied Data Mining
 in KDD ’02: Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining
"... This paper quantitatively analyzes indicators of Agent (policy seller), Adjuster (indemnity claim adjuster), Producer (policy purchaser/holder) indemnity behavior suggestive of collusion in the United States Department of Agriculture (USDA) Risk Management Agency (RMA) national crop insurance progra ..."
Abstract
 Add to MetaCart
This paper quantitatively analyzes indicators of Agent (policy seller), Adjuster (indemnity claim adjuster), Producer (policy purchaser/holder) indemnity behavior suggestive of collusion in the United States Department of Agriculture (USDA) Risk Management Agency (RMA) national crop insurance program. According to guidance from the federal law and using six indicator variables of indemnity behavior, those entities equal to or exceeding 150% of the county mean (computed using a simple jackknife procedure) on all entityrelevant indicators were flagged as "anomalous." Log linear analysis was used to test (1) hierarchical nodenode arrangements and (2) a nonrecursive model of node information sharing. Chisquare distributed deviance statistic identified the optimal log linear model. The results of the applied data mining technique used here suggest that the nonrecursive triplet and Agentproducer doublet collusion probabilistically accounts for the greatest proportion of waste, fraud, and abuse in the federal crop insurance program. Triplet and Agentproducer doublets need detailed investigation for possible collusion. Hence, this data mining technique provided a high level of confidence when 24 million records were quantitatively analyzed for possible fraud, waste, or other abuse of the crop insurance program administered by the USDA RMA, and suspect entities reported to USDA. This data mining technique can be applied where vast amounts of data are available to detect patterns of collusion or conspiracy as may be of interest to the criminal justice or intelligence agencies.
Comparison of the Probabilities of Misclassification for the Estimated Linear, Quadratic, and Unbiaseddensity Discrimant Functions . . .
, 1988
"... ..."
NONPARAMETRIC ESTIMATES OF LOW BIAS Authors:
"... We consider the problem of estimating an arbitrary smooth functional of k ≥ 1 distribution functions (d.f.s) in terms of random samples from them. The natural estimate replaces the d.f.s by their empirical d.f.s. Its bias is generally ∼ n −1, where n is the minimum sample size, with a p th order ite ..."
Abstract
 Add to MetaCart
We consider the problem of estimating an arbitrary smooth functional of k ≥ 1 distribution functions (d.f.s) in terms of random samples from them. The natural estimate replaces the d.f.s by their empirical d.f.s. Its bias is generally ∼ n −1, where n is the minimum sample size, with a p th order iterative estimate of bias ∼ n −p for any p. For p ≤ 4, we give an explicit estimate in terms of the first 2p −2 von Mises derivatives of the functional evaluated at the empirical d.f.s. These may be used to obtain unbiased estimates, where these exist and are of known form in terms of the sample sizes; our form for such unbiased estimates is much simpler than that obtained using polykays and tables of the symmetric functions. Examples include functions of a mean vector (such as the ratio of two means and the inverse of a mean), standard deviation, correlation, return times and exceedances. These p th order estimates require only ∼ n calculations. This is in sharp contrast with computationally intensive bias reduction methods such as the p th order bootstrap and jackknife, which require ∼ n p calculations. KeyWords: bias reduction; correlation; exceedances; multisample; multivariate; nonparametric; ratio of means; return times; standard deviation; von Mises derivatives. AMS Subject Classification: • 62G05, 62G30. 230 C.S. Withers and S. NadarajahNonparametric Estimates of Low Bias 231