Distribution-Conditioned Adversarial Variational Autoencoder for Valid Instrumental Variable Generation

Authors: Xinshu Li, Lina Yao

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experimental results validate the effectiveness, stability and generality of our proposed model in generating valid IV factors in the absence of valid IV candidates. Experiments Datasets Simulated Dataset The Demands dataset is first constructed by (Hartford et al. 2017)... Real-World Dataset Following previous methods (Louizos et al. 2017), we conduct experiments on one real-world dataset: Twins. Baselines and Metrics We compare our VIV with the state-of-the-art IV generation methods on the IV regression backbones. Results on Demands Table 1 shows the performance of VIV on a simulated Demands-0.5 dataset, where 0.5 denotes the value of ρ in Eq. (12).
Researcher Affiliation Academia Xinshu Li1*, Lina Yao1, 2 1 School of Computer Science and Engineering, The University of New South Wales 2 CSIRO s Data 61 xinshu.li@unsw.edu.au, lina.yao@data61.csiro.au
Pseudocode Yes Due to the page limitation, we leave the pseudo codes in the supplementary materials.
Open Source Code Yes The project page with the code and the supplementary materials is available at Git Hub3. 3https://github.com/Xinshu LI2022/VIV
Open Datasets Yes Simulated Dataset The Demands dataset is first constructed by (Hartford et al. 2017), which simulates the causal effects of price variation T on airline demands Y. Real-World Dataset Following previous methods (Louizos et al. 2017), we conduct experiments on one real-world dataset: Twins.
Dataset Splits Yes We generate 10000/10000/10000 samples in the training/validation/testing dataset and perform 10 trials for each methods. Following (Li and Yao 2022), the Twins dataset is divided 56/24/20 into training/validation/testing sets.
Hardware Specification No Due to the page limitation, we leave the hardware environment used for the experiment and optimal hyper-parameters in supplementary materials. The paper states that hardware information is in supplementary materials, but it is not explicitly described in the main text.
Software Dependencies No The paper does not provide specific ancillary software details, such as library names with version numbers, needed to replicate the experiment.
Experiment Setup No Due to the page limitation, we leave the hardware environment used for the experiment and optimal hyper-parameters in supplementary materials. The paper states that optimal hyperparameters are in supplementary materials, but they are not explicitly described in the main text.