This method is a technique that approximates solutions that more closely match the purpose by modeling the evolution of living organisms that unfolds in the natural world and evolving candidate solutions.

Let's take a look at the concept of the genetic algorithm, based on an actual example from the natural world.

For instance, a group of lions will fight to create a harem. Various lions exist, with a strong lion being a circle, a weak one being an ex, and an intermediate one being a triangle.

Given that the lions fight for a harem, the probability that a strong lion will win is high. The weak lions will die.

After fighting, the strongest lion will breed the offspring of the next generation. There is a high probability that the offspring of the next generation will be relatively strong. Through this repetition, lions that can strongly adapt to the environment will survive.

Also, if the mutation takes place and the environment of harem changes, lions that have adapted to such changes will survive. Thus, the generations change to survive with exploring a broad range of possibilities. This is the evolution of the natural world.

What would happen if we were to replace the events of this natural world with an X-ray inspection system?

With an X-ray inspection system, a large number of image processing procedures would be created.

Then, processing the images of foreign objects. The inspection setting with a circle detects foreign objects well, the setting with an ex cannot detect foreign objects. X-ray inspection system selects only the setting, which can detect the foreign objects.

Same as in the example of the lions, the generations are repeated. The best inspection setting will ultimately emerge to achieve accurate detection.