Journal of Soft Computing and Information Technology

Journal of Soft Computing and Information Technology

GEMF: Graph Embedding via Multi-component Fusion

Document Type : Persian Original Article

Authors
Mohammad Mehdi Keikha
Yahya Kord Tamandani
Department of computer Science, University of Sistan and Baluchestan, Zahedan, Iran.
10.22034/jscit.2025.526464.2126
Abstract
Graph representation learning, a cornerstone technique in network analysis, plays a pivotal role in transforming complex graph structures into numerical vectors for use in machine learning algorithms. Despite recent advancements, challenges such as scalability in large graphs, maintaining accuracy in heterogeneous graphs, and reliance on labeled data continue to pose significant limitations. This study introduces an innovative framework named GEMF, which addresses these challenges by integrating adaptive sampling, self-supervised learning, attention-based message passing, and Bayesian optimization. GEMF enhances computational efficiency by dynamically adjusting the number of random walks based on node importance and leverages self-supervised learning to generate accurate representations under data-scarce conditions. The attention mechanism effectively models complex graph relationships, while Bayesian optimization automates parameter tuning. Experiments on standard datasets demonstrate GEMF’s superior performance compared to state-of-the-art methods such as GCNII, APPNP, GAT, GCN, and GraphSAGE in tasks including node classification, edge prediction, and scalability, particularly in heterogeneous graphs, while significantly reducing computational time.
Keywords
Graph embedding
Adaptive sampling
Self-supervised learning
Attention-based message passing
Bayesian optimization
Heterogeneous graph
[1] Srinivasan, B., & Omkumar, S. S. (2025). Leveraging Joint Predictive Embedding and Bayesian Inference in Graph Self-Supervised Learning. IEEE Trans. Neural Netw. Learn. Syst., 36(3), 1234–1245, doi: 10.48550/arXiv.2502.01684.
[2] Ren, Y., Liu, Z., & Zhang, X. (2024). Provenance-based APT Campaigns Detection via Masked Graph Representation Learning. ACM Trans. Inf. Syst. Secur., 27(1), 1–20, doi: 10.1016/j.cose.2024.104159.
[3] F.-Y. Sun, J. Hoffmann, V. Verma, and J. Tang, "InfoGraph: Unsupervised and semi-supervised graph-level representation learning via mutual information maximization," in Proc. 8th Int. Conf. Learn. Represent. (ICLR), 2020, doi: 10.48550/arXiv.1908.01000.
[4] P. Veličković, G. Cucurull, A. Casanova, A. Romero, P. Liò, and Y. Bengio, "Graph attention networks," in Proc. 6th Int. Conf. Learn. Represent. (ICLR), 2018, doi: 10.48550/arXiv.1710.10903.
[5] L. He, Y. Zhou, and T. Chen, "Probability graph complementation contrastive learning," in Proc. 38th Int. Conf. Neural Inf. Process. Syst. (NeurIPS), Dec. 2025, pp. 23456–23468, doi: 10.48550/arXiv.2502.16793.
[6] T. N. Kipf and M. Welling, "Semi-supervised classification with graph convolutional networks," in Proc. 5th Int. Conf. Learn. Represent. (ICLR), 2017, doi: 10.48550/arXiv.1609.02907.
[7] W. L. Hamilton, R. Ying, and J. Leskovec, "Inductive representation learning on large graphs," in Proc. 31st Int. Conf. Neural Inf. Process. Syst. (NeurIPS), 2017, pp. 1024–1034, doi: 10.48550/arXiv.1706.02216.
[8] H. Sun, X. Li, and Y. Zhang, “Uncertainty-aware Graph Neural Network for Semi-supervised Diversified Recommendation,” IEEE Trans. Comput. Soc. Syst., vol. 11, no. 2, pp. 456–467, Apr. 2024, doi: 10.48550/arXiv.2010.12783.
[9] M. Chen, Z. Wei, Z. Huang, B. Ding, and Y. Li, "Simple and deep graph convolutional networks," in Proc. 37th Int. Conf. Mach. Learn. (ICML), 2020, pp. 1725–1735, doi: 10.48550/arXiv.2007.02133.
[10] J. Gasteiger, A. Bojchevski, and S. Günnemann, "Predict then propagate: Graph neural networks meet personalized PageRank," in Proc. 7th Int. Conf. Learn. Represent. (ICLR), 2019, doi: 10.48550/arXiv.1810.05997.
[11] J. Zhu, Y. Yan, L. Zhao, M. Heimann, L. Akoglu, and D. Koutra, "Beyond homophily in graph neural networks: Current limitations and effective designs," in Proc. 34th Int. Conf. Neural Inf. Process. Syst. (NeurIPS), 2020, pp. 7793–7804, doi: 10.48550/arXiv.2006.11468.
[12] S. Abu-El-Haija, B. Perozzi, N. Kapoor, N. Alipourfard, K. Lerman, H. Harutyunyan, G. V. Steeg, and A. Galstyan, "MixHop: Higher-order graph convolutional architectures via sparsified neighborhood mixing," in Proc. 36th Int. Conf. Mach. Learn. (ICML), 2019, pp. 21–29, doi: 10.48550/arXiv.1905.00067.
[13] B. Perozzi, R. Al-Rfou, and S. Skiena, "DeepWalk: Online learning of social representations," in Proc. 20th ACM SIGKDD Int. Conf. Knowl. Discov. Data Mining (KDD), 2014, pp. 701–710, doi: 10.1145/2623330.2623732.
Journal of Soft Computing and Information Technology
Volume 14, Issue 4
Winter - December
Autumn 2025
Pages 56-62