Improving Domain Generalization with Domain Relations
Authors: Huaxiu Yao, Xinyu Yang, Xinyi Pan, Shengchao Liu, Pang Wei Koh, Chelsea Finn
ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirically, we evaluate the effectiveness of D3G using real-world datasets for tasks such as temperature regression, land use classification, and molecule-protein binding affinity prediction. Our results show that D3G consistently outperforms state-of-the-art methods. |
| Researcher Affiliation | Academia | 1Stanford University, 2UNC-Chapel Hill, 3CMU, 4UCLA, 5Caltech, 6University of Washington |
| Pseudocode | Yes | To summary, the pseudocodes of training and testing stages of D3G is detailed in Alg. 1. |
| Open Source Code | No | Code to reproduce our results will be made publicly available. |
| Open Datasets | Yes | TPT-48 is a weather prediction dataset from the nClim Div and nClim Grid (Vose et al., 2014) databases... FMoW... from the WILDS benchmark (Koh et al., 2021b)... The ChEMBL-STRING (Liu et al., 2022) dataset provides both the binding affinity score and the corresponding domain relation. Follow Liu et al. (2022) and treat proteins and pairwise relations as nodes and edges in the relation graph, respectively. |
| Dataset Splits | Yes | The number of training, validation, and test domains are all set as 5. ... For the 24 test domains in these two tasks, we further random split them into 12 validation domains and 12 test domains. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU/CPU models, memory, or cloud computing instance types used for running the experiments. |
| Software Dependencies | No | The paper mentions software components like PyTorch and specific model architectures (DenseNet-121, GIN) but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | All hyperparameters are selected via cross-validation. Detailed setups and baseline descriptions are provided in Appendix D. ... We list the hyperparameters in Table 4 for all datasets. |