Exponential Family Graph Embeddings

Authors: Abdulkadir Celikkanat, Fragkiskos D. Malliaros3357-3364

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experimental evaluation on real-world datasets demonstrates that the proposed techniques outperform well-known baseline methods in two downstream machine learning tasks.
Researcher Affiliation Academia Abdulkadir C elikkanat Centrale Sup elec and Inria Saclay University of Paris-Saclay Gif-Sur-Yvette, France abdulkadir.celikkanat@centralesupelec.fr Fragkiskos D. Malliaros Centrale Sup elec and Inria Saclay University of Paris-Saclay Gif-Sur-Yvette, France fragkiskos.malliaros@centralesupelec.fr
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks that are clearly labeled as such or formatted like code procedures.
Open Source Code Yes Source code. The implementation of the proposed models is provided in the following website: https://abdcelikkanat.github.io/projects/EFGE/.
Open Datasets Yes Table 1: Statistics of network datasets used in the experiments. |V|: number of nodes, |E|: number of edges, |K|: number of labels and |C|: number of connected components.
Dataset Splits Yes In our experiments, we split the nodes into varying training ratios, from 2% up to 90% in order to better evaluate the models.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact CPU/GPU models, memory, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper mentions using 'scikit-learn package' but does not specify its version number or any other software dependencies with version details, which are necessary for reproducibility.
Experiment Setup Yes For the optimization we use Stochastic Gradient Descent (SGD) (Bottou 1991) to learn representations Ω = (α, β)... we adopt the negative sampling strategy, setting sampling size to k = 5 in all the experiments. ... we have used walk length L = 10, number of walks N = 80 and window size γ = 10 for all models and the variants of EFGE model are fed with the same node sequences produced by NODE2VEC.