The causal interpretation of Bayesian Networks

摘要: The common interpretation of Bayesian networks is that they are vehicles for representing probability distributions, in a graphical form supportive human understanding and with computational mechanisms probabilistic reasoning (updating). But the assumed by causal discovery algorithms causal: links graphs specifically represent direct connections between variables. However, there some tension these two interpretations. philosophy causation posits particular connection two, namely relations certain kinds give rise to kinds. Causal take advantage this kind ruling out given observational data not supported posited probability-causality relation. discovered (remaining) then causal, simply arbitrary representations probability.