Mathematical model to detect anomalies using Sensitivity Analysis applying to neural network

The transition to the digitalization in various spheres of economic and social activity is accompanied by the emergence of large amounts of data, processing which it is necessary to identify certain dependencies and build models of processes or systems. The task to identify anomaly values in dig data sets is relevant. Existing algorithms to detect anomalies are based on different approaches and have their own advantages and disadvantages. However basic schemes of all methods are similar and use at the initial stage the separation of data in a typical for system or process and those that are not, then follow structural and parametric identification of the model, and at the final stage the trained model is used to separate the data. To improve the accuracy of algorithms, they can be modified to take into account the data structure or to combine heterogeneous mathematical models. The paper describes a combined approach to build the system for detecting anomalies based on the Isolation Forest algorithm and sequential application of a neural network classifier. To reduce the dimension of neural network input vector, the approach to Sensitivity Analysis based on applying Analysis of Finite Fluctuations to the neural network model is synthesized and described. It is presented the numerical example that shows the adequacy of the proposed approach to data analysis.

1. Parmar J. D., Patel J. T. Anomaly Detection in Data Mining: A Review. International Journal of Advanced Research in Computer Science and Software Engineering. (2017);7(4):32-40.

2. ternational Conference on Data Mining, ICDM (2008).

3. Liu F. T., Ting K. M., Zhou Z. H. Isolation-Based Anomaly Detection. ACM Trans. Knowl. Discov. Data. (2012).

4. Rosenwasser E. N., Yusupov R. M. Sensitivity of Control Systems. Moscow: Nauka. (1981).

5. Saltelli A., Ratto M., Andres T., Campolongo F., Cariboni J., Gatelli D., Saisana M., Tarantola S. Global Sensitivity Analysis – The Primer. John Wiley & Sons, Hoboken (2008).

6. Saltelli A., Tarantola S., Campolongo F. Sensitivity analysis as an ingredient of modeling. Stat. Sci. (2000); 15(4):377–395.

7. Kurowicka D., Cooke R. Uncertainty Analysis with High Dimensional Dependence Modelling. Wiley. (2006).

8. Cacuci D. G. Sensitivity and Uncertainty Analysis Theory. Chapman and Hall/CRC. (2005).

9. Hamby D. M. A review of techniques for parameter sensitivity analysis of environmental models. Environ. Monit. Assess. (1994); 32(2):135–154.

10. Box G. E. P., Meyer R. D. An analysis for unreplicated fractional factorials. Technometrics. (1986);28:11–18.

11. Dean A. M., Lewis S. M. Screening: Methods for Experimentation in Industry, Drug Discovery and Genetics. Springer, New York. (2006). 6).

12. Overstall A. M., Woods D. C. Multivariate emulation of computer simulators: model selection and diagnostics with application to a humanitarian relief model. J. Roy. Statist. Soc. C. (2016).

13. Bandura R. Composite indicators and rankings: inventory 2011. Tech. rep., United Nations Development Programme – Office of Development Studies.

14. Maozhun S., Ji L. Improved Garson algorithm based on neural network model. Proceedings of the 29th Chinese Control and Decision Conference, CCDC. (2017).

15. Blyumin S. L., Sukhanov V. F., Chebotarev S. V. Economic Factor Analysis. Lipetsk: LEGI Publishing. (2004).

16. Blyumin S. L., Borovkova G. S., Serova K. V., Sysoev A. S. Analysis of finite fluctuations for solving big data management problems. Application of Information and Communication Technologies (AICT), 9th International Conference. (2015).

17. Sysoev A. S., Scheglevatych R. V. Approaches to Detect Anomalies in Big Data of Medical Nature. Proceedings of International Conference HTCS2018. (2018).

18. Sysoev A., Ciurlia A., Sheglevatych R., Blyumin S. Sensitivity Analysis of Neural Network Models: Applying Methods of Analysis of Finite Fluctuations. Periodica Polytechnica Electrical Engineering and Computer Science [Internet]. (2019) [cited 2020];63(4):306-11. Available from: https://pp.bme.hu/eecs/article/view/14654