1e:T584,## Contributions - **Introduced a measure of dependence B(X , Y|D)** based on Bayesian comparison of models of independence (H0) and dependence (H1). - **Outlined properties of B(X , Y|D)** such as its behavior under varying data sizes and conditions. - **Presented practical applications** particularly in neuroscience, demonstrating the measure's utility in real-world data. ## Previous Methods and Their Limitations - **Review of existing measures of dependence** such as mutual information and correlation coefficients. - **Noted limitations** in interpreting model estimators on finite data sizes and issues with noise affecting measures of dependence. ## Our Theory and Methods - **Defined B(X , Y|D) as the posterior probability p(H1|D)**, quantifying the evidence favoring dependence in X and Y relative to independence. - **Discussed Bayesian model comparison** methods and assumptions for choosing models H0 and H1, including the consideration of nested models and copulas. - **Highlighted both theoretical underpinnings and simulation-based investigations** to analyze scenarios such as noise impact. ## Experiments Settings & Results - **Conducted simulations** evaluating B(X , Y|D) under various conditions (noise levels, dataset sizes). - **Results showed the measure's sensitivity to changes in data size and quality**, providing expected directional behavior reflecting underlying dependence.1f:T420,

notesum.ai

An inferential measure of dependence between two systems using Bayesian model comparison