Characteristic Circuits

Authors: Zhongjie Yu, Martin Trapp, Kristian Kersting

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conclude by presenting an empirical evaluation and discussion of the new model class. 5 Experimental Evaluation Our intention here is to evaluate the performance of characteristic circuit on synthetic data sets and UCI data sets, consisting of heterogeneous data.
Researcher Affiliation Academia Zhongjie Yu TU Darmstadt Darmstadt, Germany Martin Trapp Aalto University Espoo, Finland Kristian Kersting TU Darmstadt/Hessian.AI/DFKI Darmstadt, Germany
Pseudocode Yes Algorithm 1 CC Structure Learning
Open Source Code Yes 1Source code is available at https://github.com/ml-research/Characteristic Circuits
Open Datasets Yes Our intention here is to evaluate the performance of characteristic circuit on synthetic data sets and UCI data sets, consisting of heterogeneous data. We employed the heterogeneous data from the UCI data sets, see Molina et al. [2018] and Vergari et al. [2019] for more details on the data sets.
Dataset Splits Yes For both data sets MM and BN, 800 instances were generated for training and 800 for testing.
Hardware Specification No The paper does not specify any particular hardware (e.g., GPU, CPU models, or memory) used for the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup Yes min_k = 100, and k S = k P = 2. Various leaf types were evaluated: CC with ECF as leaves (CC-E), CC with normal distribution for continuous RVs and categorical distributions for discrete RVs, i.e., parametric leaves (CC-P), and CC with normal distribution for all leaf nodes (CC-N). The likelihoods were computed based on the inversion theorem. For discrete and Gaussian leaves, the likelihoods were computed analytically. For α-stable leaves, the likelihoods were computed via numerical integration using the Gauss-Hermit quadrature of degree 50.