VISUALIZATION OF GENETIC ALGORITHM BASED ON 2-D GRAPH TO ACCELERATE THE SEARCHING WITH HUMAN INTERVENTIONS.

The Genetic Algorithm is an area in the field of Artificial Intelligence that is
founded on the principles of biological evolution. Visualization techniques help in
understanding the searching behaviour of Genetic Algorithm. lt also makes possible
the user interactions during the searching process. It is noted that active user
intervention increases the acceleration of Genetic Algorithm towards an optimal
solution.
In proposed research work, the user is aided by a visualization based on the
representation of multidimensional Genetic Algorithm data on 2-0 space. The aim of
the proposed approach is to study the benefit of using visualization techniques to
explorer Genetic Algorithm data based on gene values. The user participates in the
search by proposing a new individual. This is difTerent from existing Interactive
Genetic Algorithm in which selection and evaluation of solutions is done by the users.
A tool termed as VIGA-20 (Visualization of Genetic Algorithm using 2-0 Graph) is
implemented to accomplish this goal. This visual tool enables the display of the
evolution of gene values from generation to generation to observing and analysing the
behaviour of the search space with user interactions. Individuals for the next
generation are selected by using the objective function. Hence, a novel humanmachine
interaction is developed in the proposed approach.
The efficiency of the proposed approach is evaluated by two benchmark
functions. The analysis and comparison of VIGA-20 is based on convergence test
against the results obtained from the Simple Genetic Algorithm. This comparison is
based on the same parameters except for the interactions of the user. The application
of proposed approach is the modelling the branching structures by deriving a rule
from best solution of VIGA-20. The comparison of results is based on the different
user's perceptions, their involvement in the VIGA-20 and the difference of the fitness
convergence as compared to Simple Genetic Algorithm.