Efficient Discovery of Actual Causality using Abstraction-Refinement

TL;DR

This paper introduces an abstraction-refinement method for efficiently discovering actual causes of safety violations in complex engineered systems, leveraging SMT solving to handle large datasets and models.

Contribution

It formulates actual causality discovery as an SMT problem and proposes a novel abstraction-refinement technique to improve scalability in causal analysis.

Findings

Significant speedup in causality detection across case studies

Effective identification of root causes in safety violations

Applicable to neural networks, reinforcement learning controllers, and autopilot systems

Abstract

Causality is the relationship where one event contributes to the production of another, with the cause being partly responsible for the effect and the effect partly dependent on the cause. In this paper, we propose a novel and effective method to formally reason about the causal effect of events in engineered systems, with application for finding the root-cause of safety violations in embedded and cyber-physical systems. We are motivated by the notion of actual causality by Halpern and Pearl, which focuses on the causal effect of particular events rather than type-level causality, which attempts to make general statements about scientific and natural phenomena. Our first contribution is formulating discovery of actual causality in computing systems modeled by transition systems as an SMT solving problem. Since datasets for causality analysis tend to be large, in order to tackle the scalability problem of automated formal reasoning, our second contribution is a novel technique based on abstraction-refinement that allows identifying for actual causes within smaller abstract causal models. We demonstrate the effectiveness of our approach (by several orders of magnitude) using three case studies to find the actual cause of violations of safety in (1) a neural network controller for a Mountain Car, (2) a controller for a Lunar Lander obtained by reinforcement learning, and (3) an MPC controller for an F-16 autopilot simulator.