Probabilistic Loss and its Online Characterization for Simplified Decision Making Under Uncertainty
This addresses the problem of computational efficiency in decision-making algorithms for scenarios where uncertainty is critical, though it appears incremental by building on existing simplification techniques.
The paper tackles the computational burden in decision-making under uncertainty by extending the mechanism to include all stochastic variability and introducing a framework to simplify decision-making while assessing and controlling the simplification's impact online, with verification through extensive simulations.
It is a long-standing objective to ease the computation burden incurred by the decision making process. Identification of this mechanism's sensitivity to simplification has tremendous ramifications. Yet, algorithms for decision making under uncertainty usually lean on approximations or heuristics without quantifying their effect. Therefore, challenging scenarios could severely impair the performance of such methods. In this paper, we extend the decision making mechanism to the whole by removing standard approximations and considering all previously suppressed stochastic sources of variability. On top of this extension, our key contribution is a novel framework to simplify decision making while assessing and controlling online the simplification's impact. Furthermore, we present novel stochastic bounds on the return and characterize online the effect of simplification using this framework on a particular simplification technique - reducing the number of samples in belief representation for planning. Finally, we verify the advantages of our approach through extensive simulations.