همایش بین المللی هوش مصنوعی

صفحه اصلی / دومین همایش بین المللی هوش مصنوعی

Adaptive Data Analysis for Density Estimation: A Pólya-Tree Approach

نویسندگان :

Amir Hossein Hadavi¹ Mohammad Reza Aref² Mohammad Mahdi Mojahedian³

1- Sharif University of Technology 2- Sharif University of Technology 3- Sharif University of Technology

کلمات کلیدی :

Adaptive data analysis،Bayesian density estimation،Non-parametric priors،Pólya trees

چکیده :

This paper revises the Adaptive Data Analysis (ADA) framework by modeling more constructive analyst–data interactions aimed at improving inference accuracy. Focusing on distribution estimation within a nonparametric Bayesian setting, we employ Pólya trees (PTs) as priors to enable adaptive selection of nonparametric queries that reduce estimation error without increasing query numbers. The hierarchical, interpretable, and conjugate nature of the framework allows subjective beliefs to be intuitively encoded as priors and efficiently updated to posteriors with tractable computations. Simulations validate the superior efficiency and robustness of the model compared with the non-adaptive approach. This structured ADA modeling offers a more realistic framework for research in human-in-the-loop systems and cognitive modeling of belief updating.