HIERARCHICAL GRAPH VISUALIZATION USING NEURAL NETWORKS

An algorithm based on a Hopfield network for solving the hierarchical graph visualization problem is presented. It simultaneously minimizes the number of crossings and total path length to produce two-dimension al drawings easily interpreted by human observers. Traditional heurist ics often follow a more local optimization approach where ''readabilit y'' criteria are sequentially applied, such as applying the barycentri c heuristic followed by the priority layout heuristic As a result of t he more global approach, the neural network achieved comparable crossi ng minimization to the barycentric heuristic while simultaneously redu cing total path length up to 50% over the priority layout heuristic fo r the benchmarks tested.