Incorporating Interpretable Output Constraints in Bayesian Neural Networks

Abstract

Domains where supervised models are deployed often come with task-specific constraints, such as prior expert knowledge on the ground-truth function, or desiderata like safety and fairness. We introduce a novel probabilistic framework for reasoning with such constraints and formulate a prior that enables us to effectively incorporate them into Bayesian neural networks (BNNs), including a variant that can be amortized over tasks. The resulting Output-Constrained BNN (OC-BNN) is fully consistent with the Bayesian framework for uncertainty quantification and is amenable to black-box inference. Unlike typical BNN inference in uninterpretable parameter space, OC-BNNs widen the range of functional knowledge that can be incorporated, especially for model users without expertise in machine learning. We demonstrate the efficacy of OC-BNNs on real-world datasets, spanning multiple domains such as healthcare, criminal justice, and credit scoring.

11 pages, with six supplementary pages. 34th Conference on Neural Information Processing Systems (NeurIPS 2020), Vancouver, Canada. Code available at: https://github.com/dtak/ocbnn-public. Updated version (final, official submission to NeurIPS in January 2021) includes post-conference revisions: improved results in Section 6.2, and corrected minor errata in Appendix C