A Unified Framework for Human-Allied Learning of Probabilistic Circuits

TL;DR

This paper introduces a unified framework for probabilistic circuits that integrates domain knowledge into parameter learning, enhancing performance especially in data-scarce, knowledge-rich domains like healthcare.

Contribution

It presents a novel framework that systematically incorporates domain knowledge into probabilistic circuit learning, addressing limitations of existing data-driven approaches.

Findings

Outperforms purely data-driven methods on benchmarks.

Effectively leverages domain knowledge for improved accuracy.

Demonstrates efficiency in real-world healthcare datasets.

Abstract

Probabilistic Circuits (PCs) have emerged as an efficient framework for representing and learning complex probability distributions. Nevertheless, the existing body of research on PCs predominantly concentrates on data-driven parameter learning, often neglecting the potential of knowledge-intensive learning, a particular issue in data-scarce/knowledge-rich domains such as healthcare. To bridge this gap, we propose a novel unified framework that can systematically integrate diverse domain knowledge into the parameter learning process of PCs. Experiments on several benchmarks as well as real world datasets show that our proposed framework can both effectively and efficiently leverage domain knowledge to achieve superior performance compared to purely data-driven learning approaches.