A likelihood-based deep conditional density learning method with an application to ITR.

Please see the article "Deep Conditional Distribution Learning: Non-asymptotic Theory, Inference, and an Application to Optimal Individualized Treatment Rules." My work was accomplished at the Hong Kong Polytechnic University. I will never forget this wonderful experience at HK PolyU and am deeply grateful for it. Happy New Year, may the lights of Victoria Harbour continue to shine brightly.

For CDL, please run $\textbf{main-parallel.py}$ and $\textbf{main-paralleled.py}$ for $d=5$ and $d=15$, respectively. For files Model_d5 and Model_d15, please run $\textbf{Main500.ipynb}$, $\textbf{Main1000.ipynb}$ and $\textbf{Main2000.ipynb}$ for ITR estimation, and run $\textbf{AHtest-size.ipynb}$, $\textbf{AHtest-power.ipynb}$ for the heterogeneity test.

$\textbf{Notification:}$ This project is protected by the Copyright Law of the P.R.C (link: https://www.gov.cn/guoqing/2021-10/29/content_5647633.htm) and the Copyright Ordinance (Cap. 528) (link: https://www.elegislation.gov.hk/hk/cap528). Any unauthorized copying, modification, or distribution will result in legal consequences, and the author reserves all rights to enforce its intellectual property.

Copyright © 2025 Qiyue Huang, created in Hong Kong, China. All rights reserved.

31/12/2025, Hung Hom, Kowloon, Hong Kong, China.