Beyond tree-shaped credal probabilistic circuits
Probabilistic CTL
Tree (set theory)
DOI:
10.1016/j.ijar.2023.109047
Publication Date:
2023-11-01T01:54:21Z
AUTHORS (5)
ABSTRACT
Probabilistic circuits are a class of probabilistic generative models that allow us to compute different types queries in polynomial time. Unlike many the mainstream approaches for modeling, they can exact likelihoods, marginals, and expectations. Yet, assessing reliability their inferences is not straightforward. Credal imprecise counterpart allowing, among other queries, computations cautious sensitivity analyses. In this work, we propose an efficient algorithm lower upper expectations factorizing functions using credal circuit. We discuss under what structural assumptions works. prove such has time complexity input size. general case, computing NP-hard problem, yet proposed be used as approximation. Some experiments show how approximation degrades with model structure.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (22)
CITATIONS (1)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....