Neural Causal Abstractions
Authors: Kevin Xia, Elias Bareinboim
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we show how representation learning can be used to learn abstractions, which we apply in our experiments to scale causal inferences to high dimensional settings such as with image data. and In this section, we empirically evaluate the effects of utilizing abstractions in causal inference tasks. Details of data-generating models and architectures can be found in Appendix C. Implementation code is publicly available at https://github.com/Causal AILab/Neural Causal Abstractions. |
| Researcher Affiliation | Academia | Kevin Xia, Elias Bareinboim Causal AI Lab, Columbia University {kevinmxia, eb}@cs.columbia.edu |
| Pseudocode | Yes | Algorithm 1: Constructing MH from ML. and Algorithm 2: Neural Abstract ID Identifying and estimating queries across abstractions using NCMs. |
| Open Source Code | Yes | Implementation code is publicly available at https://github.com/Causal AILab/Neural Causal Abstractions. |
| Open Datasets | Yes | We evaluate the RNCM in a high-dimensional image dataset of colorized MNIST (Deng 2012) digits. |
| Dataset Splits | No | The paper mentions 'training iterations' in Figure 6a, but it does not specify the exact percentages or counts for training, validation, or test dataset splits, nor does it reference predefined standard splits with citations for reproducibility. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running the experiments, such as GPU models, CPU types, or cloud computing specifications. |
| Software Dependencies | No | The paper mentions the use of 'Neural Causal Models' and the 'deep learning toolkit', but it does not provide specific version numbers for any software dependencies or libraries (e.g., Python, PyTorch, TensorFlow versions). |
| Experiment Setup | No | The paper states that 'Details of data-generating models and architectures can be found in Appendix C.' and 'see details in App. C.' regarding the GAN-NCM implementation. However, the main text does not explicitly provide concrete hyperparameter values or training configurations for the experimental setup. |