An algorithmic approach to identify irrelevant information in sequential teams

摘要: An algorithmic framework that identifies irrelevant data (i.e., may be ignored without any loss of optimality) at agents a sequential team is presented. This relies on capturing the properties do not depend specifics state spaces, probability law, system dynamics, or cost functions. To capture these notion form developed. A then modeled as directed acyclic graph and identified using D-separation specific subsets nodes in graph. provides an procedure for identifying ignoring agents, thereby simplifying control laws need to implemented.