Cool project to use the genetic algorithm?

Question

I am looking for a practical application for using the genetic algorithm. Some things thought of:

• Website interface optimization
• Optimize your car with a physics simulator
• Genetic programming
• Auto test case creation

But no one jumped at me. So, if you had free time (several months) to spend on the project of genetic algorithms, what would you decide to do?

Answer 1

One topic with many features is to use evolutionary algorithms to develop game strategies. People used evolution to create strategies for poker, checkers / drafts, Go, and many other games. J-GAP people used genetic programming to develop bots for Robocode .

I recently published an introductory article on evolutionary computing. It includes details of some things for which evolutionary algorithms have been used. Adam Marchik also wrote an excellent article with many examples. Argonaut's genetic blog contains dozens of links to interesting evolutionary projects.

A less common type of evolutionary algorithm is the learning classifier system . This develops a set of rules for classifying input data. It can be applied to the same problems that neural networks use. It would be interesting to develop LCS for a specific problem, for example, to try to predict sports results based on form.

Answer 2

I look at the evolving lego structures , by far the most interesting toy project for GA.

http://static.23.nu/md/Pictures/ZZ099735B6.jpg

Answer 3

You might be interested in something like Roger Alsing Mona Lisa

Answer 4

I read about something similar proposed in the issue of automating the production of test cases. In fact, it was quite the opposite. You specify test cases, and do not allow the algorithm (genetic programming) to create the code that passes. In this way, you provide specifications, and you breed your program. If you google a little for this, you can find some research on this.
I think this is a very intriguing idea. Of course, now no one will create applications like this, but this is a great area of ​​research.

Answer 5

Perhaps find a trip for sellers . I recently tried to find a walking tour of around 66 places in Paris, and I found coding all of these things a lot of fun. I thought that my excitement comes from two sources: the fact that I could draw tours (also one good thing about the idea of ​​Mona Lisa, or if you are designing your thing, Lego structures) and the fact that there are so many approaches that You can Compare:

• You can use simple heuristics, for example, "start somewhere and always go to the nearest, nondescript place."
• You can use more complex heuristics, such as "continue to add edges (anywhere) in order of increasing cost, unless it will cause three edges to exit from the same vertex or smaller cycle."
• You can use approximation algorithms such as simple, based on minimal spanning trees , if your costs satisfy the triangle inequality. As a bonus, you can encode minimal spanning trees (maybe you can even use GA there, despite having fast and simple accurate algorithms). If you feel confident, you can program a more sophisticated (3/2) -approximation algorithm based on minimal spanning trees and perfect mappings.
• If you have a tour of traveling sellers and it crosses itself, you can improve it by "not crossing it." Again, this implies some conditions for the cost function.
• You can even try to find the tour yourself and see if you can beat your computer.
• And finally, genetic algorithms! The problem is very well suited for GA, because there is a very clear fitness function and a very easy way to do recombination.

Answer 6

See What problems can be solved with genetic programming?

I thought the design of the vehicle was really cool.

Answer 7

It was used in complex synthesizers to create sound patches (for example, Clavia Nord Modular G2 ) I would try it for other Yamaha DX7 machines that comes to mind, but I think there are quite a few software alternatives. Or image generation.

Read the theory of patch algorithms in a Palle Dahlstedt article. It seems at the moment this is not so ...

Answer 8

Genetic algorithms are well suited for optimization and planning. An example would be planning a set of machines having parts and operators over time to complete a set of tasks. Although perhaps this is not the most exciting project, it will have real-world applications.

Answer 9

In my opinion, I saw a mention of the project , to use it to try to find the optimal keyboard layout, it would be - beat Dvorak, as it were: D

Answer 10

I had fun playing with Conway's life game and genetic algorithms,
trying to develop advanced life forms. Not sure if this will keep you busy for months, though ..

Answer 11

Returning to college, I performed a multidimensional minimization of functions. Say you have af (x) that takes the parameters x1, x2, x3, ..., xn and generates the value Y .. you need to find the parameters x1, .., xn such that Y = Y1 .. not so complicated .. an interesting way to find out though. Although Nedlermead is more effective ... it does not tend to get stuck in local lows.

Answer 12

Create musical compositions! Read about David Cope and his program Emily Howell.