CodeMorph uses advanced machine learning techniques to evolve functions which fit training data sets using an easy-to-use interface. CodeMorph devises mathematical models from data with a minimal amount of configuration. Its takes as input a set of training data and a grammar which describes the rules whereby the output functions are formed. Multi-variable and multi-output functions are fully supported. Various parameters allow for controlling the evolutionary process such as the solution population, solution depth/complexity, mutation rate, selection criteria, and more. The plotting panel shows the progress of the evolutionary process as it runs. CodeMorph takes advantage of multiple processors/cores to push your machine to its full potential. The system uses MML (Meta Modeling Language), a full-featured programming language to allow not only functions but larger blocks of code to be generated and allows the mathematical portions of this code to be simplified. CodeMorph 2.0 introduces a new project type Code Generation for generating code embedded within a larger MML program. This is a flexible system whereby the MML program can use the generated code in any way it chooses and programmatically determines its error metric which is passed back to the evolutionary process.
If you have questions or comments, please contact us at:
Copyright 2011 AndroMeta LLC.
... Learn more about CodeMorph by reading the CodeMorph User’s Guide [PDF].
