DEMO-TDQL: An adaptive multi-objective optimization algorithm

Author

Rakshit, Pratyusha ; Konar, Amit ; Eunjin Kim ; Nagar, Atulya K.

Author_Institution

ETCE Dept., Jadavpur Univ., Kolkata, India

fYear

2013

fDate

20-23 June 2013

Firstpage

3095

Lastpage

3102

Abstract

An adaptive memetic algorithm incorporates an adaptive selection of memes (units of cultural transmission) from a meme-pool to improve the cultural characteristics of the individual member of a population-based search algorithm. The paper proposes an extension of Multi-objective Optimization realized with Differential Evolution algorithm by utilizing the composite benefits of Differential Evolution for Multi-objective Optimization (DEMO) for global search and Templocal refinementoral Difference Q-Learning (TDQL) for local refinement. Computer simulations performed on a well known set of 23 benchmark functions reveal that the proposed algorithm outperforms its competitors with respect to inverted generational distance, spacing and error ratio.

Keywords

evolutionary computation; learning (artificial intelligence); optimisation; search problems; DEMO-TDQL; adaptive memetic algorithm; adaptive multiobjective optimization algorithm; benchmark functions; differential evolution for multiobjective optimization; error ratio; global search; individual member cultural characteristics; inverted generational distance; local refinement; meme-pool; memes adaptive selection; population-based search algorithm; spacing ratio; temporal difference q-learning; Algorithm design and analysis; Benchmark testing; Measurement; Optimization; Sociology; Statistics; Vectors; Q-learning; adaptive memetic algorithm; differential evolution; multiobjective optimization; non-dominance;

fLanguage

English

Publisher

ieee

Conference_Titel

Evolutionary Computation (CEC), 2013 IEEE Congress on

Conference_Location

Cancun

Print_ISBN

978-1-4799-0453-2

Electronic_ISBN

978-1-4799-0452-5

Type

conf

DOI

10.1109/CEC.2013.6557947

Filename

6557947