Original title: Genetic Algorithms for Tractable Bayesian Network Fusion via Pre-Fusion Edge Pruning
Pablo Torrijos, José A. Gámez, José M. Puerta, Juan A. Aledo · September 2026
Combining several Bayesian networks—models that map out connections between variables—into one network is hard. Merging everything keeps every connection, but makes the new network too complex to calculate answers quickly. Cutting random connections keeps calculations fast, but it can drop vital links or keep noise. Researchers designed a consensus framework to solve this. It focuses on connections that appear across multiple networks while capping complexity before the merge happens. They built genetic algorithmsOptimization algorithms inspired by natural selection, using mechanisms like mutation and crossover to explore solutions.—problem-solving programs modeled on natural selection—using advanced starting setups, customized modification steps, and a tailored scoring function. Testing on synthetic and real-world networks showed these genetic algorithms work better than adapted existing tools and simpler greedy baseline methods.
Imagine writing a group report where each member has their own draft. If you combine all drafts, the report becomes too long and confusing. If you only use sentences everyone agrees on, you might miss key points. In this paper, merging Bayesian networks is handled by a system that picks the common parts while keeping the final report concise and effective.
Bayesian networks model uncertainty in fields like medicine or finance, but combining them can make models too complex to compute or lose important relationships. This method helps create practical, accurate models from multiple sources without sacrificing scalability.
The abstract does not specify the computational cost of the genetic algorithms or how they scale with very large networks. It also does not address failure cases or performance when input networks are highly divergent.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Genetic Algorithms for Tractable Bayesian Network Fusion via Pre-Fusion Edge Pruning — Pablo Torrijos, José A. Gámez, José M. Puerta et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.