GeneXproTools Environment

Evolving a Model
 
In order to evolve a model with GeneXproTools 4.0 you must first choose the problem category which is chosen while you are loading the input data for modeling. There are four different types of problem category in GeneXproTools 4.0: Function Finding, Classification, Time Series Prediction, and Logic Synthesis, each using specific learning algorithms and requiring special features.

To Evolve a Model with GeneXproTools 4.0
  1. In the Settings Panel, select the General Settings Tab and then choose the appropriate settings for your problem.
    For all types of problems, the most important is the chromosome architecture, that is, the number of genes, the head size and the linking function.
  2. In the Functions Panel, choose the appropriate set of functions to model your data.
    GeneXproTools 4.0 offers a great variety of built-in functions (a total of 279 for mathematical models and 258 for Boolean problems). Besides, GeneXproTools 4.0 also allows you to create both static and dynamic user-defined functions and evolve complex models with them. Static UDFs allow you to establish a permanent relationship between specific independent variables, for instance UDF1 = do + d1 + d2 + d3, and use this new UDF as an extra variable/terminal in your modeling kit. Dynamic UDFs are quite different from static UDFs as they behave exactly like the built-in functions of GeneXproTools and therefore can be used to model all kinds of relationships between variables or complex expressions. For instance, you can design DDF1 so that it will model the sum of four variables or expressions, that is, DDF1 = (expression 1) + (expression 2) + (expression 3) + (expression 4), where the value of each expression will depend on the context of DDF1 in the expression tree. A note of caution, though, although extremely interesting, DDFs decrease the speed of the algorithm somewhat and therefore we advise you to choose your functions from the wide set of GeneXproTools 4.0 built-in functions.
  3. In the Settings Panel, select the Fitness Function Tab and choose the fitness function.
    GeneXproTools 4.0 provides you with 36 built-in fitness functions for Function Finding and Time Series Prediction problems and 30 different fitness functions for Classification and Logic Synthesis problems. In addition, GeneXproTools 4.0 also allows you to create your own fitness functions and therefore allows you to explore even more solution landscapes.
  4. In the Settings Panel, select the Genetic Operators Tab and choose the appropriate degree of genetic variation.
    GeneXproTools 4.0 offers several genetic operators which allow you to introduce the ideal degree of genetic modification in order to allow an efficient evolution of your models. If you are overwhelmed by all these operators, please use the defaults as they work very well for all problems.
  5. In the Run Panel, click the Evolve button.
    GeneXproTools 4.0 shows all the essential features of the best-of-generation model during the discovery process, including a Curve Fitting chart and a Heat Map of the variables that are being used by the best models. Whenever you are satisfied with the result, just press the Stop button. Your best-of-run model is ready either for analysis or scoring. And if you are still not happy with the results, you can still try to fine-tune the evolved model by pressing the Optimize button or design a simpler version of it by pressing the Simplify button.


Home | Contents | Previous  | Next