Flexible and Approximate Computation through State-Space Reduction

In the real world, insufficient information, limited computation resources, and com­ plex problem structures often force an au­ tonomous agent to make a decision in time less than that required to solve the prob­ lem at hand completely. Flexible and ap­ proximate computation are two approaches to decision making under limited computa­ tion resources. Flexible computation helps an agent to flexibly allocate limited compu­ tation resources so that the overall system utility is maximized. Approximate compu­ tation enables an agent to find the best sat­ isfactory solution within a deadline. In this paper, we present two state-space reduction methods for flexible and approximate compu­ tation: quantitative reduction to deal with inaccurate heuristic information, and struc­ tural reduction to handle complex problem structures. These two methods can be ap­ plied successively to continuously improve so­ lution quality if more computation is avail­ able. Our results show that these reduction methods are effective and efficient, finding better solutions with less computation than some existing well-known methods.