Learning from What's Right and Learning from What's Wrong

Bart Jacobs
(Institute for Computing and Information Sciences (iCIS), Radboud University Nijmegen, The Netherlands)

The concept of updating (or conditioning or revising) a probability distribution is fundamental in (machine) learning and in predictive coding theory. The two main approaches for doing so are called Pearl's rule and Jeffrey's rule. Here we make, for the first time, mathematically precise what distinguishes them: Pearl's rule increases validity (expected value) and Jeffrey's rule decreases (Kullback-Leibler) divergence. This forms an instance of a more general distinction between learning from what's right and learning from what's wrong. The difference between these two approaches is illustrated in a mock cognitive scenario.

In Ana Sokolova: Proceedings 37th Conference on Mathematical Foundations of Programming Semantics (MFPS 2021), Hybrid: Salzburg, Austria and Online, 30th August - 2nd September, 2021, Electronic Proceedings in Theoretical Computer Science 351, pp. 116–133.
Published: 29th December 2021.

ArXived at: https://dx.doi.org/10.4204/EPTCS.351.8 bibtex PDF

Comments and questions to: eptcs@eptcs.org
For website issues: webmaster@eptcs.org