ADVANTAGES OF GENETIC ALGORITHM

The advantages of genetic algorithm includes,

1    Parallelism and liability

2.   Many ways to speed up and improve a GA application as knowledge about problem domain is gained

3.   Solution space is wider

4.   The fitness landscape is complex

5.    Easy to discover global optimum

6.    The problem has multi objective function

7.    Only uses function evaluations.

8     Easily modified for different problems.

9.    Handles noisy functions well.

10.  Handles large, poorly understood search spaces easily

11.  Good for multi-modal problems returns a suite of solutions.

12.  Very robust to difficulties in the evaluation of the objective function. [1]

13.  They require no knowledge or gradient information about the response surface

14.  Discontinuities present on the response surface have little effect on overall optimization performance

15.  They perform very well for large-scale optimization problems

16.  Can be employed for a wide variety of optimization problems [2]

17.  Genetic algorithm is a method which is very easy to understand and it practically does not demand the knowledge of mathematics.[3]

PREMATURE CONVERGENCE IN GENETIC ALGORITHM

GA is a kind of hill-climbing search; more specifically it is very similar to a randomized beam search. As with all hill-climbing algorithms, there is a problem of local maxima. There is no absolute assurance that a genetic algorithm will find a global optimum. It happens very often when the populations have a lot of subjects. Local maxima in a genetic problem are those individuals that get stuck with a pretty good, but not optimal, fitness measure. [7]

When applying genetic algorithm for solving large-scale and complex real-world problems, premature convergence is the one of the most frequently encountered difficulties. In that situation  , the  solving  procedure  is  trapped  in  the suboptimal state and  most  of the operators can’t  produce offspring surpassing their parents any more . It has been proven that the genetic algorithm can't converge to the global optimal solution [8]

Any small mutation gives worse fitness. Fortunately, crossover can help them get out of a local maximum. Also, mutation is a random process, so it is possible that we may have a sudden large mutation to get these individuals out of this situation. (In fact, these individuals never get out. It's their offspring that get out of local maxima.) Overall, GAs have less problems with local maxima than back-propagation neural networks. [9]

Sources:

[7] [9]  C. R. Dyer, CS 540 Lecture Notes, University of Wisconsin – Madison

[8] Rudolph G., 1994, “Convergence Analysis of Canonical Genetic Algorithms”, IEEE Trans.  On Neural Networks, Vo1.5(1), p96-101.