GraphReach: Position-Aware Graph Neural Network using Reachability Estimations

Authors: Sunil Nishad, Shubhangi Agarwal, Arnab Bhattacharya, Sayan Ranu

IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical evaluation against state-of-the-art GNN architectures reveal that GRAPHREACH provides up to 40% relative improvement in accuracy. In addition, it is more robust to adversarial attacks. 4 Experiments In this section, we benchmark GRAPHREACH and establish that GRAPHREACH provides up to 40% relative improvement over state-of-the-art GNN architectures.
Researcher Affiliation Academia Sunil Nishad1 , Shubhangi Agarwal1 , Arnab Bhattacharya1 and Sayan Ranu2 1Indian Institute of Technology Kanpur, India 2Indian Institute of Technology Delhi, India snishad@cse.iitk.ac.in, sagarwal@cse.iitk.ac.in, arnabb@cse.iitk.ac.in, sayanranu@cse.iitd.ac.in
Pseudocode Yes Algorithm 1 GRAPHREACH Input: Graph G = (V, E, X, W); Anchors {ai}; Message computation function F; Message aggregation function S; Number of layers L; Non-linear function σ Output: Node embedding zv, v V 1: h0 v xv, v V 2: for l = 1, , L do 3: for v V do 4: for i = 1, , k do 5: c M l v[i] F(v, ai, hl 1 v , hl 1 ai ) Msg Computation 6: Ml v = ( ai A c M l v[i]) Wl M Concatenation: Eq. (5) 7: hl v S Ml v Msg Aggr: Eq. (6) and Eq. (8) 8: return zv Rk σ(ML v .WZ), v V
Open Source Code Yes The implementation is available at https://github.com/idea-iitd/Graph Reach.
Open Datasets Yes Datasets: We evaluate GRAPHREACH on the datasets listed in Table 2. Further details are available in App. E. Table 2: Characteristics of graph datasets used.
Dataset Splits No The paper mentions 'inductive and transductive settings' and 'unseen test graph', but does not explicitly provide specific train/validation/test dataset split percentages or absolute counts in the provided text.
Hardware Specification No The paper does not provide specific hardware details such as GPU/CPU models, processor types, or memory amounts used for running its experiments.
Software Dependencies No The paper mentions parameters and refers to Appendix E for details, but the provided text does not specify software names with version numbers (e.g., Python 3.x, PyTorch 1.x, TensorFlow 2.x) that are needed to replicate the experiment.
Experiment Setup Yes Unless specifically mentioned, we set the number of anchors (k) as log2 n. The length of each random walk, lw, is set to graph diameter, and the number of walks nw as 50. We also conducted experiments to analyze how these parameters influence prediction accuracy of GRAPHREACH.