ISSN 0869-6632 (Print)
ISSN 2542-1905 (Online)


For citation:

Sutyagin A. A., Kanakov O. I. Collective classifier learning strategy based upon competition in the coexistence regime. Izvestiya VUZ. Applied Nonlinear Dynamics, 2021, vol. 29, iss. 2, pp. 220-239. DOI: 10.18500/0869-6632-2021-29-2-220-239

This is an open access article distributed under the terms of Creative Commons Attribution 4.0 International License (CC-BY 4.0).
Full text:
(downloads: 416)
Language: 
Russian
Article type: 
Article
UDC: 
530.182

Collective classifier learning strategy based upon competition in the coexistence regime

Autors: 
Sutyagin A. A., Federal state budgetary educational institution of higher professional education Nizhny Novgorod state University named N. And.Lobachevsky
Kanakov Oleg Igorevich, Federal state budgetary educational institution of higher professional education Nizhny Novgorod state University named N. And.Lobachevsky
Abstract: 

The purpose of this research is to create a new learning strategy for collective classifiers aimed at approximating the Bayesian optimal classification rule. A collective classifier is an ensemble of simple elements, each characterized by a specific response function and free of internal dynamics or variable parameters. Learning is achieved by targeted reshaping the composition of the ensemble (quantities of elements of specific types). A formerly known strategy was based on winnertake-all competitive dynamics, which limited dramatically the capabilities of the learning strategy – e.g. in application to classes with bimodal probability distribution. Methods. The population dynamics of the learning strategy was modified to ensure a globally stable coexistence regime. The system evolution is described by a discrete stochastic selection algorithm for the ensemble elements and in the form of ordinary differential equations as a continuous approximation. An analytic expression for the stable equilibrium is available, thus allowing targeted algorithm design for achieving a prescribed learning outcome. A learning algorithm designed for approximating the Bayesian optimal classification rule is described in the paper. Results. By numerical integration of differential equations and by simulating the selection algorithm for a sample classification task with a bimodal probability distribution we confirm all theoretical statements, including the mutual agreement between the discrete and continuous descriptions, the coexistence of elements with different response functions in the trained ensemble, and achieving a summary ensemble output approximating the Bayesian classifier. The simulation suggests that relative fluctuations of the discrete variables (quantities of elements of specific types) may be lowered by increasing the total ensemble size. Conclusion. The collective classifier principle implies its possible use in creating classifiers from ensembles of simple elements, such as «smart dust» which means miniature extremely simplified sensor devices, or genetically reprogrammed living cell populations acting as biosensors. Achieving the coexistence regime in the learning strategy complicates its implementation, but broadens the range of amenable classification tasks.

Reference: 
  1. Aivazyan SA, Buchstaber VM, Yenyukov IS, Meshalkin LD. Applied Statistics: Classification and Reduction of Dimensionality. Moscow: Finance and statistics; 1989. 608 p. (in Russian).
  2. Alpaydin E. Introduction to Machine Learning. Cambridge, Massachusetts: MIT press; 2020. 712 p.
  3. Gmurman VE. Probability Theory and Mathematical Statistics. Moscow: URAIT; 2019. 479 p. (in Russian).
  4. Jameson SC. Maintaining the norm: T-cell homeostasis. Nature Reviews Immunology. 2002; 2(8):547–556. DOI: 10.1038/nri853.
  5. Starr TK, Jameson SC, Hogquist KA. Positive and negative selection of T cells. Annual Review of Immunology. 2003;21(1):139–176. DOI: 10.1146/annurev.immunol.21.120601.141107.
  6. Palmer E. The T-cell antigen receptor: A logical response to an unknown ligand. Journal of Receptors and Signal Transduction. 2006;26(5–6):367–378. DOI: 10.1080/10799890600919094.
  7. Surh CD, Sprent J. Homeostasis of naive and memory T cells. Immunity. 2008;29(6):848–862. DOI: 10.1016/j.immuni.2008.11.002.
  8. Ivanchenko MV. Competition in the twocomponent model of the immune T-cell ensemble. Izvestiya VUZ. Applied Nonlinear Dynamics. 2010;18(3):33–45 (in Russian). DOI: 10.18500/0869-6632-2010-18-3-33-45.
  9. Ivanchenko MV. Transient selection in multicellular immune networks. JETP Letters. 2011;93(1): 35–40. DOI: 10.1134/S0021364011010048.
  10. Bolkhovskaya OV, Zorin DY, Ivanchenko MV. Assessing T cell clonal size distribution: a non-parametric approach. PLoS One. 2014;9(10):e108658. DOI: 10.1371/journal.pone.0108658.
  11. Dasgupta D, Yu S, Nino F. Recent advances in artificial immune systems: models and applications. Applied Soft Computing. 2011;11(2):1574–1587. DOI: 10.1016/j.asoc.2010.08.024.
  12. Didovyk A, Kanakov OI, Ivanchenko MV et al. Distributed classifier based on genetically engineered bacterial cell cultures. ACS Synthetic Biology. 2015;4(1):72–82. DOI: 10.1021/sb500235p.
  13. Kanakov O, Kotelnikov R, Alsaedi A et al. Multi-input distributed classifiers for synthetic genetic circuits. PLoS One. 2015;10(5):e0125144. DOI: 10.1371/journal.pone.0125144.
  14. Schaerli Y, Isalan M. Building synthetic gene circuits from combinatorial libraries: screening and selection strategies. Molecular BioSystems. 2013;9(7):1559–1567. DOI: 10.1039/c2mb25483b.
  15. Sailor MJ, Link JR. «Smart dust»: nanostructured devices in a grain of sand. Chemical Communications. 2005;(11):1375–1383. DOI: 10.1039/b417554a.
  16. Trubetskov DI. Phenomenon of Lotka–Volterra mathematical model and similar models. Izvestiya VUZ. Applied Nonlinear Dynamics. 2011;19(2):69–88 (in Russian). DOI: 10.18500/0869-6632-2011-19-2-69-88.
  17. Rabinovich MI., Trubetskov DI. Introduction to theory of oscillations and waves. Moscow–Izhevsk: Regular and chaotic dynamics; 2000. P. 343–346 (in Russian).
  18. Gauze GF. The Struggle for Existence. Baltimore: The Williams & Wilkins company; 1934. 188 p. DOI: 10.5962/bhl.title.4489.
  19. Razzhevaikin VN. The multicomponent Gause principle in models of biological communities. Biology Bulletin Reviews. 2018;8(5):421–430. DOI: 10.1134/S2079086418050067.
  20. Goh BS. Global stability in many-species systems. The American Naturalist. 1977;111(977): 135–143. DOI: 10.1086/283144.
  21. Dickschat JS. Quorum sensing and bacterial biofilms. Natural Product Reports. 2010;27(3): 343–369. DOI: 10.1039/b804469b.
Received: 
10.08.2020
Accepted: 
11.11.2020
Published: 
31.03.2021