Selective Ensembles for Consistent Predictions
This addresses inconsistency issues in high-stakes applications, offering a method to ensure reliable predictions and attributions, though it is incremental as it builds on ensemble techniques.
The paper tackles the problem of inconsistent predictions and feature attributions among models trained to the same objective, which is undesirable in high-stakes contexts like medical diagnosis and finance. It introduces selective ensembles that use hypothesis testing to achieve consistent outcomes, empirically demonstrating zero inconsistent predictions on benchmark datasets with abstention rates as low as 1.5%.
Recent work has shown that models trained to the same objective, and which achieve similar measures of accuracy on consistent test data, may nonetheless behave very differently on individual predictions. This inconsistency is undesirable in high-stakes contexts, such as medical diagnosis and finance. We show that this inconsistent behavior extends beyond predictions to feature attributions, which may likewise have negative implications for the intelligibility of a model, and one's ability to find recourse for subjects. We then introduce selective ensembles to mitigate such inconsistencies by applying hypothesis testing to the predictions of a set of models trained using randomly-selected starting conditions; importantly, selective ensembles can abstain in cases where a consistent outcome cannot be achieved up to a specified confidence level. We prove that that prediction disagreement between selective ensembles is bounded, and empirically demonstrate that selective ensembles achieve consistent predictions and feature attributions while maintaining low abstention rates. On several benchmark datasets, selective ensembles reach zero inconsistently predicted points, with abstention rates as low 1.5%.