Supplemental Case Acquisition Using Mixed-Initiative Control

摘要: Learning by observation allows a software agent to learn watching an expert perform task. This transfers the burden of training from expert, who would traditionally need program agent, itself. Most existing approaches learning their in purely passive manner. We propose case-based reasoning that is able observe passively but can also use mixed-initiative control request assistance for difficult input problems. Our uses case acquisition game Tetris. show obtain cases it not have been with alone, improve its performance and places less on expert.