Bayesian Inference Seminar

Intersubjective agreements in contexts of uncertainty

curso


Unlike the formal sciences, which validate their propositions within closed axiomatic systems, the empirical sciences (from physics to the social sciences) must validate their propositions in open systems that by definition always contain some degree of uncertainty. Is it possible to reach “truths” if it is unavoidable to say “I don’t know”. Yes. The strict application of the rules of probability (Bayesian approach) guarantees intersubjective agreements in contexts of uncertainty, the foundation of empirical truths. Under uncertainty the logic is paraconsistent insofar as it is necessary to believe at the same time in A and not A until the surprise, the only source of information, decides. Because this selection process is, like evolutionary, multiplicative in nature (a single zero in the sequence of reproduction and survival generates an extinction), there is an advantage in favor of variants that reduce fluctuations. Although the strict application of probability theory has been shown to be the ideal logic in uncertainty contexts, its adoption was historically limited due to the high computational cost associated with it: Unlike the frequentist approach to probability that selects a single hypothesis, the Bayesian approach updates the beliefs of each and every hypothesis according to empirical and formal evidence (data and models). Although computational limitations have been largely overcome in recent decades thanks to the development of efficient approximation methods, historical inertia is now their main limitation.

Avatar
Gustavo Landfried
Bayesian Data Scientist

Empirical knowledge emerges as life does