Visualizing Directed Graphs Using a Position Assignment Approach Based on Hopfield Neural Networks

One of the challenges in computer design automation is to visualize scheduling and allocation graphs. This visualization incorporates some approaches of drawing aesthetics which involve combinatorial optimization. In this paper, a method of vertex position assignment using a Hopfield-type neural network is explored to optimize directed graph visualizations. This method allows crossing number and path length minimization to be processed simultaneously. The neural network designed and implemented for this purpose has shown promising results. In addition, this method is directly applicable to any problem that requires visualization of a multi-level, directed graph. Performance results for moderate-sized directed graphs are presented. Figure 1: Same digraph, different maps and readability.