Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety

TL;DR

This survey reviews practical methods for addressing safety concerns in deep neural networks used in critical applications, categorizing their shortcomings and discussing techniques for detection, quantification, and mitigation.

Contribution

It provides a comprehensive overview of safety-related methods for DNNs, categorizing insufficiencies and discussing research efforts to address them, aiding both ML experts and safety engineers.

Findings

Broad coverage of safety techniques for DNNs

Categorization of model insufficiencies

Discussion on research directions and limitations

Abstract

The use of deep neural networks (DNNs) in safety-critical applications like mobile health and autonomous driving is challenging due to numerous model-inherent shortcomings. These shortcomings are diverse and range from a lack of generalization over insufficient interpretability to problems with malicious inputs. Cyber-physical systems employing DNNs are therefore likely to suffer from safety concerns. In recent years, a zoo of state-of-the-art techniques aiming to address these safety concerns has emerged. This work provides a structured and broad overview of them. We first identify categories of insufficiencies to then describe research activities aiming at their detection, quantification, or mitigation. Our paper addresses both machine learning experts and safety engineers: The former ones might profit from the broad range of machine learning topics covered and discussions on limitations of recent methods. The latter ones might gain insights into the specifics of modern ML methods. We moreover hope that our contribution fuels discussions on desiderata for ML systems and strategies on how to propel existing approaches accordingly.