论文信息 - Using the Delta Test for Variable Selection

Using the Delta Test for Variable Selection

Input selection is an important consideration in all large-scale modelling problems. We propose that using an established noise variance estimator known as the Delta test as the target to minimise can provide an effective input selection methodology. Theoretical justifications and experimental results are presented.

Michel Verleysen | Amaury Lendasse | Francesco Corona | Emil Eirola | Elia Liitiäinen

[1] D. François. High-dimensional data analysis : optimal metrics and feature selection/ , 2007 .

[2] Michel Verleysen,et al. On the Effects of Dimensionality on Data Analysis with Neural Networks , 2009, IWANN.

[3] Carsten Peterson,et al. Finding the Embedding Dimension and Variable Dependencies in Time Series , 1994, Neural Computation.

[4] Roberto Battiti,et al. Using mutual information for selecting features in supervised neural net learning , 1994, IEEE Trans. Neural Networks.

[5] Arup Bose,et al. Variance estimation in high dimensional regression models , 2000 .

[6] Amaury Lendasse,et al. Nearest Neighbor Distributions and Noise Variance Estimation , 2007, ESANN.

[7] Michel Verleysen,et al. The Concentration of Fractional Distances , 2007, IEEE Transactions on Knowledge and Data Engineering.

[8] Antonia J. Jones,et al. New tools in non-linear modelling and prediction , 2004, Comput. Manag. Sci..

[9] Amaury Lendasse,et al. Variable Scaling for Time Series Prediction: Application to the ESTSP'07 and the NN3 Forecasting Competitions , 2007, 2007 International Joint Conference on Neural Networks.

[10] Dafydd Evans. Data-derived estimates of noise for unknown smooth models using near-neighbour asymptotics , 2002 .