Abstract
Machine learning systems deployed in safety-critical robotics settings mustbe robust to distribution shifts. However, system designers must understand thecause of a distribution shift in order to implement the appropriateintervention or mitigation strategy and prevent system failure. In this paper,we present a novel framework for diagnosing distribution shifts in a streamingfashion by deploying multiple stochastic martingales simultaneously. We showthat knowledge of the underlying cause of a distribution shift can lead toproper interventions over the lifecycle of a deployed system. Our experimentalframework can easily be adapted to different types of distribution shifts,models, and datasets. We find that our method outperforms existing work ondiagnosing distribution shifts in terms of speed, accuracy, and flexibility,and validate the efficiency of our model in both simulated and live hardwaresettings.