Стохастична оптимизация в помощ на хипотетико-дедуктивния метод.

Abstract

This paper outlines an approach to enhance the process of scientific creativity by partial automation of the Received View type of reasoning. The idea is to implement stochastic optimization for testing hypotheses involving complex analytical models. When ample empirical data are available, it becomes possible to reach conclusions of high statistical significance about elements of a theoretical model comprising the core of a hypothesis. We illustrate the approach with two examples from computational neuroscience, showing both a success and a problematic outcome due to existing unresolved issues.