TY - GEN

T1 - New entropy model for extraction of structural information from XCS population

AU - Park, Won Kyung

AU - Oh, Jae C.

PY - 2009/12/31

Y1 - 2009/12/31

N2 - We show that when XCS is applied to complex real-valued problems, the XCS populations contain structural information. This information exists in the underlying classifier space as the degree of uncertainty associated to the problem space. Therefore, we can use structural information to improve the overall system performance. We take an information theoretic approach, introducing a new entropy model for XCS to extract the structural information from dynamically forming substructures. Using this entropy model, we can collectively emphasize or de-emphasize the effect of an individual input. For a complex problem domain, we chose the speaker identification (SID) problem. The SID problem challenges XCS with a complex problem space that may force the learning classifier system to evolve large and highly overlapping population. The entropy model improved the system performance up to 5-10% in large-set SID problems. Furthermore, the entropy model has the ability to assist the population initialization in the beginning of the learning process by assuring a certain level of overall diversity.

AB - We show that when XCS is applied to complex real-valued problems, the XCS populations contain structural information. This information exists in the underlying classifier space as the degree of uncertainty associated to the problem space. Therefore, we can use structural information to improve the overall system performance. We take an information theoretic approach, introducing a new entropy model for XCS to extract the structural information from dynamically forming substructures. Using this entropy model, we can collectively emphasize or de-emphasize the effect of an individual input. For a complex problem domain, we chose the speaker identification (SID) problem. The SID problem challenges XCS with a complex problem space that may force the learning classifier system to evolve large and highly overlapping population. The entropy model improved the system performance up to 5-10% in large-set SID problems. Furthermore, the entropy model has the ability to assist the population initialization in the beginning of the learning process by assuring a certain level of overall diversity.

KW - Information theory

KW - Learning classifier systems

KW - Speaker identification

UR - http://www.scopus.com/inward/record.url?scp=72749101974&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=72749101974&partnerID=8YFLogxK

U2 - 10.1145/1569901.1570073

DO - 10.1145/1569901.1570073

M3 - Conference contribution

AN - SCOPUS:72749101974

SN - 9781605583259

T3 - Proceedings of the 11th Annual Genetic and Evolutionary Computation Conference, GECCO-2009

SP - 1283

EP - 1290

BT - Proceedings of the 11th Annual Genetic and Evolutionary Computation Conference, GECCO-2009

T2 - 11th Annual Genetic and Evolutionary Computation Conference, GECCO-2009

Y2 - 8 July 2009 through 12 July 2009

ER -