Causal Network
P(A|B) = P(B|A)·P(A) / P(B)
Evidence Inputs
Set observations
Each node has a background probability λ₀ — the chance it activates with no active parents — modeling unknown/unobserved causes. Each directed edge carries a causal strength λᵢ: the independent probability that an active parent triggers the child node.
Inference is exact: the engine enumerates all 27 = 128 joint states, computes their probabilities using the noisy-OR CPTs, and marginalizes. When you set evidence, the posterior is computed via likelihood weighting — consistent with Bayes' theorem.
| Node | Background λ₀ | Posterior P(true) | Parent Edges |
|---|