Sensitivity Analysis in Artificial Neural Network and it's Applications on the Research of Attributes Correlation

Abstract:

Sensitivity analysis method can evaluate the importance of model input attributes. A multivariable sensitivity analysis method based on neural network connection weights and a calculation method of attributes correlation are proposed in this paper, and are applied to the research of attributes correlation. To verify the effectiveness of the proposed methods, this study employed a man-made example and a UCI-IRIS dataset to test the performance of the method. The results show that the sensitivity analysis method can really identify important and strong correlation attributes of model, and can simplify the model effectively, and can improve the accuracy of the model.

• Periodical: Advanced Materials Research (Volumes 785-786)
• Pages: 1441-1446
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.785-786.1441
• Online since: September 2013
• Authors: Hong Yan Lin*, Hai Hua Xing
• Keywords: Artificial Neural Network (ANN), Attributes Correlation, Connection Weight, Sensitivity Analysis

References

[1] CAI Yi, XING Yan, HU Dan. On Sensitivity Analysis[J]. Journal of Beijing Normal University : Natural Science, 2008 , 44(1): 9-16.

[2] Borgonovo E , Peccati L. Uncertainty and global sensitivity analysis in the evaluation of investment project s[J ] . J of Production Economics , 2006 , 104 : 62.

DOI: 10.1016/j.ijpe.2005.05.024

[3] Sun Jing. Nonlinear sensitivity analysis in investment decision-making[J]. Statistics and Decision, 2005, ( 11).

[4] Muriel Gevrey, Ioannis Dimopoulos, SovanLek. Two-way interaction of input variables in the sensitivity analysis of neural network models [ J ] . Ecological Modelling , 2006 , 195 : 43.

DOI: 10.1016/j.ecolmodel.2005.11.008

[5] XU Chonggang , HU Yuanman, etc. Sensitivity analysis in ecological modeling[J ] . CHINESE JOURNAL OF APPL IED ECOLOGY, J un. 2004 , 15 (6): 1056～1062.

[6] Chen Taicong, Han Dajian. Neural network method i n parameter sensitivity analysis and its application i n engineering[J ] . Chinese Journal of Computational Mechanics , 2004 , 21 (6) : 752.

[7] Xia Yuanyou，Xiong Haifeng. Sensibility Analysis of Slope Stability Based on Artificial Neural Network[J ] . Chinese Journal of Rock Mechanics and Engineering, 2004 , 23 (16) : 2703.

[8] JIANG Yong, QIU Rong, CHEN Jun, etc. Role of Sensitivity Analysis in the Study on Reaction Mechanism of Toxicants and Radicals in Fire[J ] . Fire Safety Science, 2002 , 11 (1) : 001.

[9] Garson G D. Interpreting neural2network connection weight s[J ] . AI Expert , 1991 , 6 (4) : 47.

[10] Tchaban T , Taylor M J , Griffin A. Establishing impact s of the input s in a feedforward neural network[J ] . Neural Compute Appl , 1998 , 7 : 309.

[11] Dimopoulos Y, Bourret P , Lek S. Use of some sensitivity criteria for choosing networks with good generalization ability[J ] . Neural Process Let t , 1995 , 2 : 1.

DOI: 10.1007/bf02309007

[12] Ruck D W , Rogers S K, Kabrisky M. Feature selection using multi layer percept rons [ J ] . Journal of Neural Network Computing , 1990 , 2 : 40.

[13] Julian D , Olden Donald A. J ackson : Illuminating the black box, : a randomization approach for understanding variable contributions in artificial neural networks[ J ] . Ecological Modelling, 2002 , 154 : 135.

DOI: 10.1016/s0304-3800(02)00064-9

[14] Scardi M , Harding. Developing an empirical model of phytoplankton primary productions : a neural network case study[J ] . Ecol Model , 1999 , 120 : 213.

DOI: 10.1016/s0304-3800(99)00103-9

[15] Muriel Gevreya , Ioannis Dimopoulosb , Sovan Leka. Two2way interaction of input variables in the sensitivity analysis of neural network models [ J ] . Ecological Modelling , 2006 , 195 : 43.