ANN model of subdivision error based on genetic algorithm

According to the test data of subdivision errors in the measuring cycle of angular measuring system, the characteristics of subdivision errors generated by this system are analyzed. It is found that the subdivision errors are mainly due to the rotary-type inductosyn itself. For the characteristic of cyclical change, the subdivision errors in other measuring cycles can be compensated by the subdivision error model in one measuring cycle. Using the measured error data as training samples, combining GA and BP algorithm, an ANN model of subdivision error is designed. Simulation results indicate that GA reduces the uncertainty in the training process of the ANN model, and enhances the generalization of the model. Compared with the error model based on the least-mean-squared method, the designed ANN model of subdivision errors can achieve higher compensating precision.

Variable Selection in High-Dimensional Error-in-Variables Models via Controlling the False Discovery Proportion

Multiple testing has gained much attention in high-dimensional statistical theory and applications,and the problem of variable selection can be regarded as a generalization of the multiple testing.It is aiming to select the important variables among many variables.Performing variable selection in high-dimensional linear models with measurement errors is challenging.Both the influence of high-dimensional parameters and measurement errors need to be considered to avoid severely biases.We consider the problem of variable selection in error-in-variables and introduce the DCoCoLasso-FDP procedure,a new variable selection method.By constructing the consistent estimator of false discovery proportion(FDP)and false discovery rate(FDR),our method can prioritize the important variables and control FDP and FDR at a specifical level in error-in-variables models.An extensive simulation study is conducted to compare DCoCoLasso-FDP procedure with existing methods in various settings,and numerical results are provided to present the efficiency of our method.

Alternative Distributions to Estimate Usual Intake of Nutrients for Groups

When assessing food intake patterns in groups of individuals, a major problem is finding usual intake distribution. This study aimed at searching for a probability distribution to estimate the usual intake of nutrients using data from a cross-sectional investigation on nutrition students from a public university in Sao Paulo state, Brazil. Data on 119 women aged 19 to 30 years old were used. All women answered a questionnaire about their lifestyle, diet and demographics. Food intake was evaluated from a non-consecutive three-day 24-hour food record. Different probability distributions were tested for vitamins C and E, panthotenic acid, folate, zinc, copper and calcium where data normalization was not possible. Empirical comparisons were performed, and inadequacy prevalence was calculated by comparing with the NRC method. It was concluded that if a more realistic distribution for usual intake is found, results can be more accurate as compared to those achieved by other methods.

Use of signal quality measurements to gain efficiency in the analysis of cDNA microarray data

This research provides a new way to measure error in microarray data in order to improve gene expression analysis. Microarray data contains many sources of error. In order to glean information about mRNA expression levels, the true signal must first be segregated from noise. This research focuses on the variation that can be captured at the spot level in cDNA microarray images. Variation at other levels, due to differences at the array, dye, and block levels, can be corrected for by a variety of existing normalization procedures. Two signal quality estimates that capture the reliability of each spot printed on a microarray are described. A parametric estimate of within-spot vari ance, referred to here as σ^2spot, assumes that pixels follow a normal distribution and are spatially correlated. A non-parametric estimate of error, called the mean square prediction error （MSPE）, assumes that spots of high quality possess pixels that are similar to their neighbors. This paper will provide a framework to use either spot quality measure in downstream analysis, specifically as weights in regression models. Using these spot quality estimates as weights can result in greater efficiency, in a statistical sense, when modeling mi- croarray data.