When static modeling fails
A static Bayesian network assumes the structure is the same at every moment in time. For many engineering and risk-management problems this is fine — the question is about steady-state behavior. For others — sepsis management, equipment aging, market regime change — the structure itself evolves, and the model must capture that evolution as a first-class property.
Pages in this cluster
Dynamic Bayesian Networks → A static Bayesian network models a system at one point in time. A dynamic Bayesian network models how that system evolves.
Stress testing, early warning indicators, and scenario simulation all require temporal structure — a model of how today’s state produces tomorrow’s. Dynamic Bayesian networks provide that structure without abandoning causal reasoning.
Gated Bayesian Networks → Your model has one set of parameters. Your organization operates in multiple regimes.
A standard Bayesian network is calibrated on a single distribution — the one that prevailed in the training data. When the regime changes — when the market moves from trending to mean-reverting, when a system shifts from normal operation to degraded state, when a counterparty moves from solvent to distressed — the conditional probability tables that were correct in the old regime are wrong in the new one. A Gated Bayesian Network encodes this explicitly: multiple models, explicit regime boundaries, automatic switching when evidence crosses a threshold.
Two different extensions
A DBN models continuous evolution: the next state is a probabilistic function of the previous one. A gated BN models discrete regime switches: the same variables behave under different structural rules in different operating modes. The pages describe each in turn.