Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
MIRACLE: Causally-Aware Imputation via Learning Missing Data Mechanisms
Authors: Trent Kyono, Yao Zhang, Alexis Bellot, Mihaela van der Schaar
NeurIPS 2021 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct extensive experiments on synthetic and a variety of publicly available datasets to show that MIRACLE is able to consistently improve imputation over a variety of benchmark methods across all three missingness scenarios: at random, completely at random, and not at random. |
| Researcher Affiliation | Academia | Trent Kyono University of California, Los Angeles EMAIL Yao Zhang University of Cambridge EMAIL Alexis Bellot University of Oxford Oxford, United Kingdom EMAIL Mihaela van der Schaar University of Cambridge University of California, Los Angeles The Alan Turing Institute EMAIL |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Source code at https://github.com/vanderschaarlab/MIRACLE. |
| Open Datasets | Yes | variety of publicly available UCI datasets [8] |
| Dataset Splits | No | The paper states "We use an 80-20 train-test split" but does not specify a validation set split. |
| Hardware Specification | No | The paper does not provide specific hardware details (like GPU/CPU models, memory, or specific computing environments) used for running its experiments. |
| Software Dependencies | No | The paper mentions "tensorflow" and other libraries like "scikit-learn" but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | We performed a hyperparameter sweep (log-based) for β1 and β2 with ranges between 1e-3 and 100. By default we have β1 and β2 set to 0.1 and 1, respectively. |