Abstract | ||
---|---|---|
In this work a new method is proposed to make simple multiple and independent classifiers, based on the behaviour of certain class of ants. In these works, the objective is to construct different clusters from the data, being the actions of the agents (ants), those that produce the global result. In this work, the same model, but with a different objective, will be followed: the agents are going to be classifiers specialized in small zones of the dominion of the data, defined by means of a prototype of that zone. The result is a set of classifiers that cover the total of the dominion of the data, and whose use is based on the election of the prototype (classifier) closest to the data to classify, either in the selection of the neighbours closest to the data, and a mechanism of combination of the results. |
Year | Venue | Keywords |
---|---|---|
2003 | IC-AI | classifier agents.,swarm intelligence,- multiple classifier system,ant |
Field | DocType | Citations |
Swarm behaviour,Computer science,Swarm intelligence,Multi-swarm optimization,Artificial intelligence,Machine learning | Conference | 0 |
PageRank | References | Authors |
0.34 | 4 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ester Del Castillo | 1 | 5 | 1.14 |
David Cerrillo | 2 | 0 | 0.34 |
Luis Jiménez | 3 | 12 | 3.17 |
Miguel Delgado | 4 | 106 | 7.81 |
José A. Olivas | 5 | 106 | 20.85 |