Domain Adaptation by Using Causal Inference to Predict Invariant Conditional Distributions
Authors: Sara Magliacane, Thijs van Ommen, Tom Claassen, Stephan Bongers, Philip Versteeg, Joris M. Mooij
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We perform an evaluation on both synthetic data and a real-world dataset based on a causal inference challenge. Figure 3a shows a boxplot of the L2 loss of the predicted Y values with respect to the true values for both the baseline and our method |
| Researcher Affiliation | Collaboration | Sara Magliacane MIT-IBM Watson AI Lab, IBM Research sara.magliacane@gmail.com Thijs van Ommen University of Amsterdam thijsvanommen@gmail.com Tom Claassen Radboud University Nijmegen tomc@cs.ru.nl |
| Pseudocode | No | The paper describes an algorithm in prose within Section 2.5 but does not present it as structured pseudocode or a clearly labeled algorithm block. |
| Open Source Code | Yes | The full source code of our implementation and the experiments is available online at https://github.com/caus-am/dom_adapt. |
| Open Datasets | Yes | The latter dataset consists of hematology-related measurements from the International Mouse Phenotyping Consortium (IMPC), which collects measurements of phenotypes of mice with different single-gene knockouts. Part of the CRM workshop on Statistical Causal Inference and Applications to Genetics, Montreal, Canada (2016). See also http://www.crm.umontreal.ca/2016/Genetics16/competition_e.php |
| Dataset Splits | No | The paper mentions using 'out-of-bag score' for feature selection, which is an internal validation method of Random Forests, but it does not provide explicit training, validation, or test dataset splits (e.g., percentages or counts) for reproduction. |
| Hardware Specification | No | The paper mentions 'our approach can handle about seven variables on a laptop computer' in the discussion, but it does not provide specific hardware details (like GPU/CPU models, memory, or detailed computer specifications) used for running the experiments. |
| Software Dependencies | Yes | We use the ASP solver clingo 4.5.4 [Gebser et al., 2014]. |
| Experiment Setup | Yes | First, we score all possible subsets of features by their out-of-bag score using the implementation of Random Forest Regressor from scikit-learn [Pedregosa et al., 2011] with default parameters. We provide as inputs the independence test results from a partial correlation test with significance level α = 0.05 and combine it with the weighting scheme from Magliacane et al. [2016]. |