- ホーム
- > 洋書
- > 英文書
- > Computer / General
Full Description
Bridging the gap between information theory and statistical inference, this work formalizes the equivalence between maximum likelihood and minimum mutual information. It establishes an invariant Pythagorean law and defines new rules for orthogonal mutual information decomposition. These methodological breakthroughs enable a rigorous two-step likelihood-ratio test for conditional independence and introduce novel mutual information identities. Ultimately, the framework refines variable and model selection procedures within Generalized Linear Models. By offering scalable methodologies for large-scale inference, this text is an essential resource for data analysts navigating the complexities of the modern AI era.



