GENETIC ALGORITHMS

A genetic Algorithm is an artificial intelligence system that mimics the evolutionary, survival-of-the-fittest process to generate increasingly better solutions to a problem. In other words, a genetic algorithm is an optimizing system: It finds the combination of inputs that give the best outputs.

So, instead of a pencil, paper, and calculator, you might use a genetic algorithm. You could input the appropriate information on the stocks, including the number of years the company has been in business, the performance of the stock over the last five years, price to earnings ratios, and other information.

You would also have to tell the genetic algorithm your exact “success” criteria. For example, you might use a growth rate in the company over the last year of at least 10 percent, a presence in the market place going back at least three years, a connection to the computer industry, and so forth. The genetic algorithm would simply combine and recombine stocks eliminating any combinations that don’t fit your criteria and continuing to the next iteration with the acceptable combinations- those that give an aggregate growth rate of at least 7.5 % while aiming for as high a growth rate as possible.

Genetic Algorithms use three concepts of evolution:

1. Selection – or survival of the fittest. The key to selection is to give preference to better outcomes.
2. Crossover – or combining portions of good outcomes in the hope of creating an even better outcome.
3. Mutation – or randomly trying combinations and evaluating the success ( or failure ) of the outcome.

Genetic algorithms are best suited to decision-making environments in which thousands, or perhaps millions, of solutions are possible. It can also find and evaluate solutions intelligently and can get through many more possibilities more thoroughly and faster than a human can. As you might imagine, businesses face decision-making environments for all sorts of problems like engineering design, computer graphics, strategies for game playing, anything, in fact, that requires optimization techniques.



Here are some other examples.

• Genetic algorithm are used by business executives to help them decide which combination of project a firm should invest in, taking complicated tax consideration into account
• They’re used by investment companies to help the trading choice and decisions.
• In any garment that you buy, the fabric alone accounts by between 35 percent and 40 percent of the selling price. So when cutting out the fabric to make the garment it’s important that there be a little waste as possible. Genetic algorithms are used to solve the problem of laying out the pieces of garment and cutting fabric in a way that leaves as little waste as possible.
• US West uses a genetic algorithm to determine the optimal configuration of fiber-optical cable in a network that may include 100,000 connection points by using selection, crossover, and mutation, the genetic algorithm can generate and evaluate millions of cable. At US West, this process used to take an experienced design almost two months. US West algorithm can solve a problem in two days, and save the company $1 million to $10 million each time it’s used.

Genetic Algorithm are good for this types of problems because they are use selection, crossover, and mutation as methods of exploring countless solutions and the respective worth of each.

You have to tell the genetic algorithm what constitutes a “good” solution. That could be low cost, high return, among other factors, since many potential solutions are useless or absurd if you create genetic algorithm to make bread, for example, it might try to boil flour to create moistness. That obviously wont work so genetic algorithm would simple throw that solution and try something else. Other solutions would eventually be good, and some of them would even be wonderful. According to David Goldbert, a genetic algorithm pioneer at the university of Illinois at Urban Campaign, evolution is the oldest and most powerful algorithm there is, “and three million years of evolution can’t be wrong!”