Multi-relational Poincaré Graph Embeddings
Authors: Ivana Balazevic, Carl Allen, Timothy Hospedales
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on the hierarchical WN18RR knowledge graph show that our Poincaré embeddings outperform their Euclidean counterpart and existing embedding methods on the link prediction task, particularly at lower dimensionality. |
| Researcher Affiliation | Collaboration | Ivana Balaževi c1 Carl Allen1 Timothy Hospedales1,2 1 School of Informatics, University of Edinburgh, UK 2 Samsung AI Centre, Cambridge, UK {ivana.balazevic, carl.allen, t.hospedales}@ed.ac.uk |
| Pseudocode | No | The paper does not contain a structured pseudocode or algorithm block. |
| Open Source Code | Yes | We implement both models in Py Torch and make our code, as well as all the subsets of the NELL-995 dataset, publicly available.2 https://github.com/ibalazevic/multirelational-poincare |
| Open Datasets | Yes | We implement both models in Py Torch and make our code, as well as all the subsets of the NELL-995 dataset, publicly available.2 https://github.com/ibalazevic/multirelational-poincare |
| Dataset Splits | No | The paper mentions using a 'validation set' but does not provide specific details on the dataset splits (e.g., exact percentages or sample counts for training, validation, and test sets). |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running experiments. |
| Software Dependencies | No | The paper mentions implementing models in 'Py Torch' but does not specify a version number for this or any other software dependency. |
| Experiment Setup | Yes | We choose the learning rate from {1, 5, 10, 20, 50, 100} by MRR on the validation set and find that the best learning rate is 50 for WN18RR and 10 for FB15k-237 for both models. (...) We set the batch size to 128 and the number of negative samples to 50. In all experiments, we set the curvature of Mu RP to c=1, since preliminary experiments showed that any material change reduced performance. |