D>
BSTU-IIT: Practice Course
Brest State Technical University, Course "Contemporary Intelligent Information Technology (CIIT)"
From students' works (2015)
All One Problem
Start with random 100 chromosomes each with 1000 genes - An example of result ... By Elena Leonidovna
Ratio of "1" vs "0" 20% (Left), 50% (Center) and 80% (Right)
100 genes seem to be too big to converge to all one chromosome!
Another example ... By Maksim Aleksandrovich
He started with 500 chromosomes, and found easy to converge to ALL ONE!
Lucky Dog
12 dogs start at (500,500) looking for a sausage at (200,800) symbolized +
An example ... By Kontsevich Bladislav
1st and 2nd generation
5th and 6th generation
20th and 500th generation when dogs finally reach the sausage!
All dogs found the sausage!
Other example ... By Kuznetsov Aleksandr
The other example of ONLY ONE dog who uses a HILLCLIMBING algorithm ... By Kozeko Elena
TSP
By a brute force: By calculating the distnacies all the possible routes ... By Medvedskij Konstantin
Each of the above 3 he produced cities at random, calculated distance matrix,
and calculated distance of all the possible routes to pick up the shortest.
E.g., for 5 cities:
But due to an exponential exprosion, we could continue this method for the nuber of cities = 7, 8, 9, 10, ...
So Medvedskij tried evolution for 10 cities. The result was successful.
Fitness vs generation was:
y=sin^6(5*pi*x) with (i) Fitness Sharing or (ii) Crowding algorithm
An example ... By Kuznetsov Aleksandr
Fitness vs generation was:
Lucky dog with 4 sausages sharing fitness
An example ... By Medvedskij Konstantin
Unfortunately, it seems not successful! But why?