Incorporating Interpretable Output Constraints in Bayesian Neural Networks

TL;DR

This paper presents a new probabilistic framework for integrating task-specific constraints into Bayesian neural networks, enhancing their interpretability and applicability across various real-world domains.

Contribution

It introduces Output-Constrained BNNs that incorporate constraints directly into the Bayesian framework, including a variant for amortized multi-task learning.

Findings

OC-BNNs effectively incorporate domain constraints.

Demonstrated success on healthcare, justice, and credit datasets.

Improves interpretability and usability of BNNs in practice.

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.