Learning Non-Explicit Control Parameters of Self-Organizing Systems

Controlling self-organizing systems with given control parameters is often unintuitive and inefficient for the user. Typically, there is some emergent behavior that the user would like to produce that is not directly controllable. Our goal is to develop an autonomous and domain-independent learning framework for adding non-explicit control parameters to self-organizing systems that provide users with more intuitive real-time control of the system. These additional controls are created by using regression to learn a mapping between the explicit control parameters and the non-explicit control parameters.