DIGNet: Learning Decomposed Patterns in Representation Balancing for Treatment Effect Estimation

Authors: Yiyan HUANG, WANG Siyi, Cheuk Hang LEUNG, Qi WU, Dongdong WANG, Zhixiang Huang

TMLR 2024 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The ablation studies verify the effectiveness of PDIG and PPBR in improving treatment effect estimation, and experimental results on benchmark datasets demonstrate the superior performance of our DIGNet model compared to baseline models.
Researcher Affiliation Collaboration Yiyan Huang EMAIL Department of Applied Mathematics, The Hong Kong Polytechnic University Siyi Wang * EMAIL School of Data Science, City University of Hong Kong Cheuk Hang Leung EMAIL School of Data Science, City University of Hong Kong Qi Wu EMAIL School of Data Science, City University of Hong Kong Dongdong Wang EMAIL JD Digits Zhixiang Huang EMAIL JD Digits
Pseudocode No The paper describes objective functions and network architectures but does not present a structured pseudocode or algorithm block.
Open Source Code No The paper includes a link to Open Review (https: // openreview. net/ forum? id= Z20FInf Wlm) for review purposes, but this is not a link to source code for the methodology described in the paper. No explicit statement of code release or repository link is provided.
Open Datasets Yes The IHDP dataset, introduced by Hill (2011), originates from the Infant Health and Development Program (IHDP). ... The potential outcomes were generated using setting A in the NPCI package Dorie (2021). ... We also conduct an additional experiments on another benchmark dataset Twins. The details and results are deferred to Section A.5
Dataset Splits Yes For each γ, we repeat the above data generating process to generate 30 different datasets, with each dataset split by the ratio of 56%/24%/20% as training/validation/test sets. ... The final IHDP dataset consists of 747 samples, comprising 139 treated samples and 608 controlled samples. The potential outcomes were generated using setting A in the NPCI package Dorie (2021). We use the same 1000 datasets as used in Shalit et al. (2017), with each dataset split by the ratio of 63%/27%/10% as training/validation/test sets.
Hardware Specification Yes All the experiments are run on Dell 7920 with one 16-core Intel Xeon Gold 6250 3.90GHz CPU and three NVIDIA Quadro RTX 6000 GPUs.
Software Dependencies No The paper mentions the use of the NPCI package Dorie (2021) but does not provide specific version numbers for any software libraries or frameworks used (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes Table 8: Hyperparameters of different models in simulation studies. ... Table 9: Hyperparameters of different models in IHDP experiments.