Academic Journal
ON KAMADA-KAWAI GRAPH LAYOUT WITH GRAPH NEURAL NETWORKS.
| Τίτλος: | ON KAMADA-KAWAI GRAPH LAYOUT WITH GRAPH NEURAL NETWORKS. |
|---|---|
| Alternate Title: | ПРО УКЛАДАННЯ ГРАФІВ АЛГОРИТМОМ КАМАДА-КАВАЇ ЗА ДОПОМОГОЮ ГРАФОВИХ НЕЙРОННИХ МЕРЕЖ (Ukrainian) |
| Συγγραφείς: | LINNYK, Olena, POLYAKOVA, Lyudmyla, ZARETSKA, Iryna |
| Πηγή: | Radioelectronic & Computer Systems / Radìoelektronnì ì Komp'ûternì Sistemi; 2025, Issue 3, p101-110, 10p |
| Θεματικοί όροι: | Graph neural networks, Visualization, Symmetry, Mathematical optimization, Cost functions |
| Περίληψη: | Traditional force-directed layout algorithms, such as the Kamada-Kawai (KK) method, are widely used for graph visualization due to their ability to produce aesthetically pleasing layouts. However, these algorithms can be computationally intensive for large graphs. The subject matter of the study is graph layout. The aim of this research is to explore the application of graph neural networks (GNNs) to improve the KK algorithm for graph layouts, resulting in a novel hybrid approach named KKNN. The key tasks addressed in this study include: 1. Development of the KKNN layout algorithm by integrating GNN-based reparameterization from the NeuLay-2 approach with the KK algorithm. 2. Evaluation of computational efficiency by comparing the computational performance of KKNN with both the original KK algorithm and the NeuLay-2. 3. Assessment of layout quality, particularly by examining the symmetry preservation, energy minimization, and aesthetic criteria such as minimal edge crossings and balanced node placement. 4. Testing on various graph types, including both random and highly structured (symmetric) graphs. The methods used are: GNN-based layout reparameterization, inspired by NeuLay-2; Kamada-Kawai graph layout algorithm; performance metrics, including time-toconvergence, energy minimization, and symmetry preservation. Our experiments demonstrate the following results: 1. KKNN converges to the energy minimum faster and achieves a lower energy state compared to the original KK. 2. KKNN not only reduces computational time but also better preserves graph symmetries compared to NeuLay-2. Conclusions. This study underscores the potential of integrating neural networks with traditional graph layout algorithms, presenting a promising approach for efficient and high-quality graph visualization. KKNN not only enhances computational performance but also ensures visually interpretable layouts. This hybrid approach offers a pathway for future research in graph visualization, where combining deep learning techniques with classical algorithms may open new possibilities for handling complex, large-scale graphs in a visually coherent and computationally efficient manner. [ABSTRACT FROM AUTHOR] |
| Copyright of Radioelectronic & Computer Systems / Radìoelektronnì ì Komp'ûternì Sistemi is the property of National Aerospace University named after M.E. Zhukovsky "Kharkiv Aviation Institute" and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Βάση Δεδομένων: | Complementary Index |
καταχωρήστε σχόλιο πρώτοι!