Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

LOHA: Direct Graph Spectral Contrastive Learning Between Low-Pass and High-Pass Views

Authors: Ziyun Zou, Yinghui Jiang, Lian Shen, Juan Liu, Xiangrong Liu

AAAI 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we present a comprehensive evaluation of LOHA, through a series of node classification experiments on 9 real-world datasets. Comparison between LOHA with other baselines and ablation studies validate the effectiveness of LOHA and help us to gain further insights.
Researcher Affiliation Academia Ziyun Zou1, Yinghui Jiang2, Lian Shen1, Juan Liu3, Xiangrong Liu1,2* 1Department of Computer Science and Technology, Xiamen University, 2 National Institute for Data Science in Health and Medicine, Xiamen University, 3Pen-Tung Sah Institute of Micro-Nano Science and Technology, Xiamen University, EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes methods through mathematical equations and textual descriptions, but does not contain a dedicated pseudocode or algorithm block.
Open Source Code No The paper does not provide an explicit statement regarding the release of source code or a link to a code repository.
Open Datasets Yes We choose widely used real-world datasets with different homophily levels to evaluate the performance of LOHA. (1) Homophilic Graphs: Cora, Citeseer, and Pub Med from (Yang, Cohen, and Salakhudinov 2016). (2) Heterophilic Graphs: Cornell, Texas, Actor and Wisconsin from (Pei et al. 2020); Chameleon from (Rozemberczki, Allen, and Sarkar 2021); Amazon-ratings(Amazon for short in tables) from (Platonov et al. 2023).
Dataset Splits Yes We follow the training and validation strategies as (Chien et al. 2021), where nodes are randomly split into 60%, 20%, and 20%. All comparative methods share the same fixed random splits.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup No Output embedding size and hyper-parameters in stage 2 are also fixed for fair comparison. More detailed settings can be found in Appendix.