Latent Space Explanation by Intervention
Authors: Itai Gat, Guy Lorberbom, Idan Schwartz, Tamir Hazan679-687
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate the effectiveness of our approach on Celeb A, where we show various visualizations for bias in the data and suggest different interventions to reveal and change bias. |
| Researcher Affiliation | Collaboration | 1 Technion Israel Institute of Technology 2 Net App |
| Pseudocode | No | The paper does not include any pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | No | The paper does not provide any statement or link indicating that source code for the methodology is openly available. |
| Open Datasets | Yes | We performed our experiments on the Celeb A dataset (Liu et al. 2015), which is annotated with 40 binary attributes. |
| Dataset Splits | No | The paper mentions 'The data splits are not necessarily balanced' in the context of label distribution, but does not explicitly provide details (percentages, counts, or predefined citations) for training, validation, and test dataset splits needed for reproduction. While it discusses accuracy of reconstructed representations, it doesn't specify a validation split. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments (e.g., GPU models, CPU types, memory). |
| Software Dependencies | No | The paper mentions using 'mean squared error loss', 'binary cross-entropy', 'Re LU activation function', and 'Dropout' but does not specify any software libraries or frameworks with their version numbers. |
| Experiment Setup | Yes | For the reconstruction loss of the DVAE, we used mean squared error loss, and for the reconstruction of the explanatory network, we used binary cross-entropy. In all networks, we used the Re LU activation function and Dropout. We used the same architectures for all the tasks. |