Probabilistic Reasoning in Intelligent Systems Quotes

“Probabilities and the Logic of “Almost True”
― Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference

“The problem with monotonic logic lies not in the hardness of its truth values, but rather in its inability to process context-dependent information.”
― Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference

“Another feature we lose in going from logic to uncertainty is incrementality.”
― Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference

“invoked only if it is absolutely needed for explaining some observed or derived phenomenon, e.g., finding your home burglarized and your alarm system silent.”
― Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference

“enter your mind, in which case a two-stage inference chain is assembled, governed by two probabilistic parameters, P(False alarm) and P(Prank call). Later, when the possibility of an earthquake enters consideration, the parameter P(False alarm) undergoes a partial explication; a fragment of knowledge is brought over from the remote frame of earthquake experiences and is appended to the link Burglary → Alarm as an alternative cause or explanation. The catchall hypothesis All other causes shrinks (to exclude earthquakes), and its parameters are readjusted. The radio announcement strengthens your suspicion in the earthquake hypothesis and permits you to properly readjust your decisions without elaborating the mechanics of the pressure transducer used in the alarm system. The remote possibility of having forgotten to push the reset button will”
― Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference

“To show what is still needed, let us examine how an ideal system might reason about the burglar alarm situation of Figure 1.2. Upon receiving the phone call from your neighbor, only the burglary hypothesis is triggered; your decision whether to drive home or stay at work is made solely on the basis of the parameter P(False alarm), which summarizes all other (unexplicated) causes for an alarm sound. After a moment’s reflection, the possibility of an April Fools’ Day joke may”
― Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference