CrystalBox: Future-Based Explanations for Input-Driven Deep RL Systems

arXiv:2302.13483v4 Announce Type: replace
Abstract: We present CrystalBox, a novel, model-agnostic, posthoc explainability framework for Deep Reinforcement Learning (DRL) controllers in the large family of input-driven environments which includes computer systems. We combine the natural decomposability of reward functions in input-driven environments with the explanatory power of decomposed returns. We propose an efficient algorithm to generate future-based explanations across both discrete and continuous control environments. Using applications such as adaptive bitrate streaming and congestion control, we demonstrate CrystalBox's capability to …

abstract arxiv computer computer systems cs.lg cs.ni cs.sy deep rl eess.sy environments explainability family framework functions future model-agnostic natural novel power reinforcement reinforcement learning returns systems type