Reconstruction of control skill by machine learning

• Ivan Bratko ^[1]
  1. [1] University of Ljubljana

     University of Ljubljana

     Eslovenia

     
• Localización: CAEPIA'97: actas / coord. por Asociación Española de Inteligencia Artificial, Vicente J. Botti Navarro, 1997, ISBN 84-8498-765-5, págs. 15-16
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Controlling a complex dynamic system, such as an aircraft or a crane, requires operator’s skill acquired through experience. In this talk we are interested in the question of how to transfer such operator’s skill into an automatic controller. One approach would be to extract the skill from the operator in a dialogue fashion, whereby the operator himself would be required to describe his skill. The operator’s description would then be appropriately formalised and built into an automatic controller. The problem with this approach is that skill is subcognitive, and usually the operator can only describe it partially and approximately. Therefore operator’s descriptions are not operational in the sense of being directly translatable into an automatic controller.