Authors:
Ayman Elkasaby
;
Akram Salah
and
Ehab Elfeky
Affiliation:
Cairo University, Egypt
Keyword(s):
Genetic Programming, Multiobjective Optimization, Epsilon Dominance, Evolutionary Algorithms.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Knowledge Discovery and Information Retrieval
;
Knowledge-Based Systems
;
Methodologies and Technologies
;
Operational Research
;
Optimization
;
Stochastic Optimization
;
Symbolic Systems
Abstract:
Multi-objective optimization is currently an active area of research, due to the difficulty of obtaining diverse
and high-quality solutions quickly. Focusing on the diversity or quality aspect means deterioration of the
other, while optimizing both results in impractically long computational times. This gives rise to
approximate measures, which relax the constraints and manage to obtain good-enough results in suitable
running times. One such measure, epsilon-dominance, relaxes the criteria by which a solution dominates
another. Combining this measure with genetic programming, an evolutionary algorithm that is flexible and
can solve sophisticated problems, makes it potentially useful in solving difficult optimization problems.
Preliminary results on small problems prove the efficacy of the method and suggest its potential on
problems with more objectives.