Bandit Online Learning on Graphs via Adaptive Optimization

Authors: Peng Yang, Peilin Zhao, Xin Gao

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on benchmark graph datasets show that the proposed bandit algorithm outperforms state-of-the-art competitors, even sometimes beats the algorithms using full information label feedback.
Researcher Affiliation Collaboration King Abdullah University of Science and Technology, Saudi Arabia 2South China University of Technology, China 3Tencent AI Lab, China
Pseudocode Yes Algorithm 1 MOLG-F: Adaptive Optimization for Online Learning on Graphs with Full Label Feedback; Algorithm 2 MOLG-B: Adaptive Optimization for Online Learning on Graphs with Bandit Feedback
Open Source Code Yes Proof. The proof is provided on the website1. 1https://github.com/Young Big Bird1985/MOLG/
Open Datasets Yes We exploit 4 real-world graph datasets to evaluate all the algorithms: 1) Coauthor2 is a coauthor graph of the DBLP dataset... 2) Cora2 is a citation graph... 3) IMDB3 is an up-to-date movie dataset... 4) Pub Med4 is a graph... 2http://www.cs.umd.edu/ sen/lbc-proj/data/ 3http://www.imdb.com/ 4http://www.cs.umd.edu/projects/linqs/projects/lbc/
Dataset Splits No The paper describes online learning where data is processed sequentially, and models are updated. However, it does not specify explicit train/validation/test splits with percentages or sample counts in the traditional sense for reproducibility of data partitioning.
Hardware Specification No The paper does not provide any specific hardware details such as GPU/CPU models, processors, or memory used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies or version numbers required to replicate the experiments.
Experiment Setup Yes For both methods, we tune the parameter φ with the grid {10^-2, ..., 10}. For MOLG-B, we fix the exploration parameter ϕt = 0.05 for all t [T]. We set b = 10 for Cora and Coauthor and b = 100 for IMDB and Pub Med due to variable graph structures. Finally, we fix d = 100 for the dimension of low-rank representation.