PILOGIC
All research Conference paper · 2005 · IJCAI-05

Compiling Bayesian Networks with Local Structure

Mark Chavira · Adnan Darwiche

The paper that scaled compilation. Better logical encodings of a network's local structure cut compile times by an order of magnitude, compiled networks that had never been compiled before, and sped up online inference by orders of magnitude against jointree.

What it showed

Earlier work had reduced compiling a Bayesian network to factoring a logical encoding of it, a framework that can exploit local structure. For networks with local structure and large tables but little outright determinism, the quality of that encoding turned out to decide both how long compilation takes and how fast answers come back afterwards.

The paper reports new findings that scale the approach significantly: order-of-magnitude improvements in compile time, networks compiled successfully for the first time, and orders-of-magnitude faster online inference on some networks compared with jointree, which does not exploit local structure. The 2008 journal paper names it as one of the three papers it is based on.

Why it matters

This is where the method first cleared problems the standard approach could not, and the first of the papers that the 2008 reference brings together.

Scope

A methods paper. No product claims attach to it.

More research.

JOURNAL · 2008 On probabilistic inference by weighted model counting Overview CONFERENCE · 2008 Diagnosing Faults in Electrical Power Systems of Spacecraft and Aircraft Overview JOURNAL · 2010 Probabilistic Model-Based Diagnosis: An Electrical Power System Case Study Overview

Don’t guess.
Compute.

Models Manifest Resolve
Company Company Careers
Resources Research News Contact
Compliance Privacy Policy Terms
PILOGIC Exact AI for aerospace
and defense.
© 2026 PiLogic · pilogic.ai