Informatics and Applications
2023, Volume 17, Issue 3, pp 71-75
CLASSIFICATION BY CAUSE-AND-EFFECT RELATIONSHIPS
- A. A. Grusho
- N. A. Grusho
- M. I. Zabezhailo
- D. V. Smirnov
- E. E. Timonina
Abstract
By definition, property A in object O is the cause for the occurrence of consequence B which is available for observation in information space I if characteristics of A can generate an object in space I containing consequence B. In this case, B determinedly appears with the appearance of A. Therefore, one can consider the classification problem as calculating the consequences of the characteristics of the object where the consequences act as characteristics of the class. In this case, the characteristics of the classification object can be considered as the cause that deterministically (classification as mapping) generates consequences (characteristics of the class).
Each of the properties Ai, i = 1,..., k, is the cause of the deterministic appearance of a nonempty set of its consequences. If the number of classes is large as well as the sets of consequences of each, then the classification problem can be complex to compute due to the fact that repetitions of consequences in the sets of consequences are possible. Therefore, it is advisable to look for simplified schemes for classifying objects according to the causes for the consequences in them. For this, an apparatus of systems of various representatives can be used. In the context of the problem of classifying causes due to consequences, it is impossible to directly use F Hall's theorem on systems
of various representatives, since elements of cause-and-effect chains cannot be broken. The paper shows that the transformation of each of the same chains of cause-and-effect relationships into one common new element in the sets of consequences forms the possibility of applying the conditions of F Hall's theorem.
[+] References (7)
- Pearl, J. 2010. Causal inference. Causality: Objectives and assessment. Eds. I. Guyon, D. Janzing, and B. Scholkopf. Proceedings of machine learning research ser. Whistler, Canada. 6:39-58.
- Richens, J. G., C. M. Lee, and S. Johri. 2020. Improving the accuracy of medical diagnosis with causal machine learning. Nat. Commun. 11:3923. 9 p. doi: 10.1038/s41467- 020-17419-7.
- Zabezhailo, M. I., A. A. Grusho, N.A. Grusho, and E. E. Timonina. 2021. Podderzhka resheniya zadach diagnosticheskogo tipa [Support for solving diagnostic type problems]. Sistemy i Sredstva Informatiki - Systems and Means of Informatics 31(1):69-81. doi: 10.14357/ 08696527210106.
- Zhang, C., K. Zhang, and Y. Li. 2021. A causal view on robustness of neural networks. arXiv.org. 21 p. Available at: https://arxiv.org/abs/2005.01095v3 (accessed June 28, 2023).
- Hall, P. 1935. On representation of subsets. J. Lond. Math. Soc. 10:26-30. doi: 10.1112/JLMS/S1-10.37.26.
- Hall, M. 1967. Combinatorial theory. London: Blaisdell Pub. Co. 310 p.
- Grusho, A.A., M.I. Zabezhailo, V.V. Kulchenkov, D. V. Smirnov, E. E. Timonina, and S.Ya. Shorgin. 2023. Prichinno-sledstvennye svyazi v zadachakh analiza nenablyudaemykh protsessov [Cause-and-effect relationships in analysis of unobservable process properties]. Sistemy i Sredstva Informatiki - Systems and Means of Informatics 33(2):71-78. doi: 10.14357/08696527230207.
[+] About this article
Title
CLASSIFICATION BY CAUSE-AND-EFFECT RELATIONSHIPS
Journal
Informatics and Applications
2023, Volume 17, Issue 3, pp 71-75
Cover Date
2023-10-10
DOI
10.14357/19922264230310
Print ISSN
1992-2264
Publisher
Institute of Informatics Problems, Russian Academy of Sciences
Additional Links
Key words
cause-and-effect relationships; finite classification; searching for the properties in unobservable data
Authors
A. A. Grusho  , N. A. Grusho , M. I. Zabezhailo , D. V. Smirnov  , and E. E. Timonina
Author Affiliations
Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
Sberbank of Russia, 19 Vavilov Str., Moscow 117999, Russian Federation
|