Insights into Data through Model Behaviour: An Explainability-driven Strategy for Data Auditing for Responsible Computer Vision Applications
This addresses data auditing for responsible computer vision applications, particularly in medical domains, by providing a complementary strategy to data-driven methods, though it is incremental as it builds on existing explainability techniques.
The study tackles the problem of hidden data quality issues in medical benchmark datasets by using an explainability-driven strategy to audit data through a dummy model's behavior, discovering that these issues cause deep learning models to make predictions for wrong reasons, and leveraging insights to create high-performing models with appropriate prediction behavior.
In this study, we take a departure and explore an explainability-driven strategy to data auditing, where actionable insights into the data at hand are discovered through the eyes of quantitative explainability on the behaviour of a dummy model prototype when exposed to data. We demonstrate this strategy by auditing two popular medical benchmark datasets, and discover hidden data quality issues that lead deep learning models to make predictions for the wrong reasons. The actionable insights gained from this explainability driven data auditing strategy is then leveraged to address the discovered issues to enable the creation of high-performing deep learning models with appropriate prediction behaviour. The hope is that such an explainability-driven strategy can be complimentary to data-driven strategies to facilitate for more responsible development of machine learning algorithms for computer vision applications.