Building a system capable to acquire and organize knowledge from interaction with the outer world is the final goal of our research. Every result of such interaction should be evaluated, saved in system's memory and trigger refinements of system's structure and invariants.

In course of such system construction we shall have to answer a set of questions, for which we were unable to find answers in literature:

	How to hierarchize knowledge to support compact storage and fast access to a large volume of constantly replenished information
	How to optimize the hierarchy of knowledge as new facts become available
	How to select dependencies that should be generalized and used as invariants
	How to redirect part of system's activity to self-learning, etc.

Such a system will necessarily contain algorithms for optimization and generalization as natural and unavoidable elements. In our development Evolutionary Search plays the role of an elementary optimization algorithm. Evolutionary Regression and Evolutionary Classification play the roles of low-level generalization algorithms.