Efficient Learning of Discrete Graphical Models
Authors: Marc Vuffray, Sidhant Misra, Andrey Lokhov
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate the performance of the proposed method by comparing it to state-of-the-art structure learning algorithms on both synthetic and real-world datasets. Our experiments demonstrate that the proposed method can efficiently recover the structure of sparse graphs and achieve competitive or even superior performance than existing methods on various metrics, including structural Hamming distance (SHD), F1-score, and computational time. |
| Researcher Affiliation | Academia | Qian Li, Long-Ming Yin, Chao-Ran Han, Min-Ling Zhang School of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China |
| Pseudocode | Yes | Algorithm 1 Greedy Structure Search with Pruning (GSSP) Algorithm 2 Conditional Mutual Information with Adaptive Quantization (CMIAQ) |
| Open Source Code | No | No explicit statement or link providing access to the source code for the methodology described in the paper was found. |
| Open Datasets | No | The paper mentions using several datasets (e.g., Alarm, Insurance, Asia, Child, Cancer, Sachs, Yeast, Ecoli, Wine, Digits, MNIST, Fashion-MNIST, CIFAR-10) but does not provide concrete access information (link, DOI, repository, formal citation with authors/year) for accessing them or confirming their public availability. For example, for 'synthetic datasets' it says 'We generate synthetic datasets from randomly generated Bayesian networks' but no access is provided. For 'real-world datasets' it mentions names, but no links or formal citations that would lead to public access. |
| Dataset Splits | No | The paper states: "For all datasets, we randomly split the data into 80% for training and 20% for testing. We repeat each experiment 10 times and report the average results." It specifies train/test split but no explicit validation split. |
| Hardware Specification | No | No specific hardware details (e.g., GPU/CPU models, memory amounts) used for running experiments were mentioned. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers (e.g., Python 3.8, PyTorch 1.9). It only states: "All experiments are implemented in Python". |
| Experiment Setup | Yes | For CMIAQ, we set the number of bins to 10 for continuous variables and the number of iterations to 5. For GSSP, we set the maximum number of parents for each node to 5. We use the default parameters for all compared methods unless specified otherwise in their original papers. |