Diffusion Maps for Textual Network Embedding

Authors: Xinyuan Zhang, Yitong Li, Dinghan Shen, Lawrence Carin

NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show that the proposed approach outperforms state-of-the-art methods on the vertex-classification and link-prediction tasks.
Researcher Affiliation Academia Department of Electrical and Computer Engineering Duke University Durham, NC 27707
Pseudocode No No explicit pseudocode block or section labeled 'Algorithm' was found in the paper.
Open Source Code No The paper does not provide an explicit statement about open-source code availability nor does it include a link to a code repository.
Open Datasets Yes We conduct experiments on three real-world datasets: DBLP [28], Cora [15], and Zhihu [26].
Dataset Splits No The paper describes training and testing splits for link prediction and multi-label classification but does not explicitly mention a separate 'validation' set or its specific use for hyperparameter tuning.
Hardware Specification Yes All models are implemented in Tensorflow using a NVIDIA Titan X GPU with 12 GB memory.
Software Dependencies No The paper mentions 'TensorFlow' but does not specify its version number or any other software dependencies with version details.
Experiment Setup Yes We set the embedding of dimension d to 200 with ds and dt both equal to 100. The number of hops H is set to 4 and the importance coefficients λh s are tuned for different datasets and different tasks with λ0 > λ1 > > λH. αtt, αss, αts, and αst are set to 1, 1, 0.3 and 0.3 respectively. The number of negative samples K is set to 1 to speed up the training process.