Timezone: »

On the Safety of Interpretable Machine Learning: A Maximum Deviation Approach
Dennis Wei · Rahul Nair · Amit Dhurandhar · Kush Varshney · Elizabeth Daly · Moninder Singh

Wed Nov 30 09:00 AM -- 11:00 AM (PST) @ Hall J #322

Interpretable and explainable machine learning has seen a recent surge of interest. We focus on safety as a key motivation behind the surge and make the relationship between interpretability and safety more quantitative. Toward assessing safety, we introduce the concept of maximum deviation via an optimization problem to find the largest deviation of a supervised learning model from a reference model regarded as safe. We then show how interpretability facilitates this safety assessment. For models including decision trees, generalized linear and additive models, the maximum deviation can be computed exactly and efficiently. For tree ensembles, which are not regarded as interpretable, discrete optimization techniques can still provide informative bounds. For a broader class of piecewise Lipschitz functions, we leverage the multi-armed bandit literature to show that interpretability produces tighter (regret) bounds on the maximum deviation. We present case studies, including one on mortgage approval, to illustrate our methods and the insights about models that may be obtained from deviation maximization.

Author Information

Dennis Wei (IBM Research)
Rahul Nair (IBM Research Europe)
Amit Dhurandhar (IBM Research)
Kush Varshney (IBM Research)
Elizabeth Daly (IBM Research)
Moninder Singh (IBM Research AI)

More from the Same Authors