Title: An unsupervised classifier system using soft graph colouring

Authors: Jorge Flores-Cruz; Pedro Lara-Velázquez; Miguel A. Gutiérrez-Andrade; Sergio Gerardo De-los-Cobos-Silva; Eric Alfredo Rincón-García; Roman Anselmo Mora-Gutiérrez; Antonin Ponsich

Addresses: Electric Engineering Department, Metropolitan Autonomous University – Iztapalapa, Av. San Rafael Atlixco 186, Col. Vicentina, Del. Iztapalapa, México D.F., C.P. 09340, México ' Electric Engineering Department, Metropolitan Autonomous University – Iztapalapa, Av. San Rafael Atlixco 186, Col. Vicentina, Del. Iztapalapa, México D.F., C.P. 09340, México ' Electric Engineering Department, Metropolitan Autonomous University – Iztapalapa, Av. San Rafael Atlixco 186, Col. Vicentina, Del. Iztapalapa, México D.F., C.P. 09340, México ' Electric Engineering Department, Metropolitan Autonomous University – Iztapalapa, Av. San Rafael Atlixco 186, Col. Vicentina, Del. Iztapalapa, México D.F., C.P. 09340, México ' Systems Department, Metropolitan Autonomous University – Azcapotzalco, Av. San Pablo 180, Colonia Reynosa Tamaulipas, México D.F., C.P. 02200, México ' Systems Department, Metropolitan Autonomous University – Azcapotzalco, Av. San Pablo 180, Colonia Reynosa Tamaulipas, México D.F., C.P. 02200, México ' Systems Department, Metropolitan Autonomous University – Azcapotzalco, Av. San Pablo 180, Colonia Reynosa Tamaulipas, México D.F., C.P. 02200, México

Abstract: Unsupervised classifiers do not require previous training to achieve their task, on the contrary, they are able to propose alternative classifications that make more sense using raw data instead of human interpretation. In this article an unsupervised classifier system using the soft graph colouring model is presented. This model has the ability to deal with risk and the probability of committing Type I or II errors, for instance, in medical diagnosis, where we want to minimise the risk of a mistaken prognosis. The proposed model is evaluated using some classical instances, and the results are compared with other classifiers, given in all cases solutions as good or better than supervised classifiers.

Keywords: pattern recognition; unsupervised classification; soft graph colouring; SGC; optimisation.

DOI: 10.1504/IJBCRM.2018.094172

International Journal of Business Continuity and Risk Management, 2018 Vol.8 No.3, pp.186 - 199

Received: 28 Nov 2017
Accepted: 21 Apr 2018
Published online: 20 Aug 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article