广义信息论用于预测的评价和优化

A Generalized Information Theory for Evaluation and Optimization of Forecasting

本文提出了和K.R.Popper科学进化模型类似的广义通信模型,推导出了广义Kullback公式和广义互信息公式。新公式体现了:信息来自预测.信息的多少需要事实检验,预测精确且与事实符合,信息就多,反之信息就少,甚至为负值。文中讨论了最佳预言的选择,说明了由广义互信息公式得出的科学理论进步标准怎样和Popper的观点一致。

The paper proposes a generalized communication model that is similar to K. R. Popper's model of evolution of science, deduces a generalized Kullback equation and a generalized cross-information equation. The basic thought of generalized information theory is that information comes from forecasts. The paper discusses the selection of the best prediction, and, explains the identity of the standards of advances of scientific theories from the generalized cross-information equation with Popper's views.