Genetic Algorithm for feature selection

I am working on genetic algorithm for feature selection in Brain MRI Images. I have done the coding part but not getting the correct results. I have used 20 chromosomes of length 10 (features = 10), tournament selection for parent selection, then crossover and mutation to create a new generation.

This process is repeated 50 times. But the results are not correct. I am not getting what is the problem, whether the features I am taking are not correct or there is some mistake in my understanding of genetic algorithm.

https://drive.google.com/folderview?id=0Bys_Tre7ZR9ufmRSR0RjRjRrc0IyNF92Rm83dURleUpaeXhTLWs0R2cwVUk4clRfTjY4djQ&usp=sharing

This the link of my shared folder named project in which my whole code is there. The main code is named "newgenetic" and the rest are functions. Please help me...

[purti choudhary]

I don't know why the images are not clear on sharing...

[Geoff Hayes]

purti - you can attach your code using the paperclip button. Make sure that you press the Attach File button once you have chosen the file. Also, please describe what you mean by the results are not correct. What does this mean to you?

[purti choudhary]

Thank you Sir.. I have attached some of the files. By saying that I am not getting correct results I meant that in case of genetic algorithm, the desired result is that the fitness value should maximise with every new generation, but its not happening in my results.

Even on making 50 iterations sometimes it is giving the best fitness value at 5th iteration only..

[purti choudhary]

These are fitness values that I am getting...

fo =

Columns 1 through 9

-25.5753  -19.5838  -21.1933  -18.1975  -19.5838  -25.5753  -15.8950  -15.8950 

 -15.8950

Columns 10 through 18

-15.8950  -15.8950  -12.8992  -17.5044  -21.1933  -15.8950  -12.8992  -12.8992

-15.2018

Columns 19 through 27

-12.8992  -15.8950  -15.8950  -15.2018  -15.8950  -12.8992  -12.8992  -15.2018

-12.8992

Columns 28 through 36

-15.2018  -15.2018  -20.5001  -15.2018  -12.8992  -15.2018  -19.8070  -15.2018

-15.2018

Columns 37 through 45

-19.8070  -17.5044  -19.8070  -15.2018  -20.5001  -15.2018  -15.2018  -15.2018

-15.2018

Columns 46 through 50

-18.1975  -15.2018  -15.2018  -15.2018  -15.2018

[Geoff Hayes]

Purti - what do the fitness values mean? From your code, it appears that you have coded up your own version of the Genetic Algorithm. Have you verified that your algorithm works for other problems?

[purti choudhary]

I tried to implement a research paper in which this fitness function function which I have used was given.