Cross-Graph Learning of Multi-Relational Associations

Authors: Hanxiao Liu, Yiming Yang

ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In our experiments with a subset of DBLP publication records and an Enzyme multi-source dataset, the proposed method successfully scaled to the large cross-graph inference problem, and outperformed other representative approaches significantly.
Researcher Affiliation Academia Hanxiao Liu HANXIAOL@CS.CMU.EDU Yiming Yang YIMING@CS.CMU.EDU Carnegie Mellon University, Pittsburgh, PA 15213, USA
Pseudocode Yes Algorithm 1: Transductive Learning over Product Graph (TOP)
Open Source Code No The paper does not provide any statement about making its source code available, nor does it include a link to a code repository.
Open Datasets Yes We evaluate our method on real-world data in two different domains: the Enzyme dataset (Yamanishi et al., 2008) for compound-protein interaction and the DBLP dataset of scientific publication records.
Dataset Splits Yes For both datasets, we randomly sample one third of known interactions for training (denoted by O), one third for validation and use the remaining ones for testing.
Hardware Specification No The paper does not provide any specific details about the hardware used for running the experiments (e.g., GPU/CPU models, memory, or specific computing environments).
Software Dependencies No The paper does not specify any software dependencies with version numbers, such as programming languages, libraries, or frameworks.
Experiment Setup Yes Following (Duchi et al., 2011), we allow adaptive step sizes for each element in α. That is, in the t-th iteration we use η(t) k1,...,k J = η0 .h Pt τ=1 α (τ) k1,...,k J 2 as the step size for αk1,...,k J, where α (τ) k1,...,k J t τ=0 are historical gradients associated with αk1,...,k J and η0 is the initial step size (set to be 1).