Adaptive Modeling for Risk-Aware Decision Making

摘要

Decision making under uncertainty is a core capability of an autonomous agent. In model-based reasoning, agents reason about the effects of their own actions and the events in the environment based on the model parameters. A cornerstone for long-term autonomy with safety guarantees is risk-aware decision making. The importance of accounting for risks in AI systems is attracting growing interest (Kulic and Croft 2005; Zilberstein 2015; Amodei et al. 2016). A risk-aware model fully accounts for a known set of risks in the environment, with respect to the problem under consideration. The process of decision making using such a model is risk-aware decision making. Formulating risk-aware models is critical for robust reasoning under uncertainty, since the impact of using less accurate models may be catastrophic in extreme cases due to overly optimistic view of problems. The risk awareness in a model can be increased by improving the model fidelity and thereby, the solution quality.