Hybrid Algorithms for the Minimum-Weight Rooted Arborescence Problem

Minimum-weight arborescence problems have recently enjoyed an increased attention due to their relation to imporant problems in computer vision. A prominent example is the automated reconstruction of consistent tree structures from noisy images. In this paper, we first propose a heuristic for tackling the minimum-weight rooted arborescence problem. Moreover, we propose an ant colony optimization algorithm. Both approaches are strongly based on dynamic programming, and can therefore be regarded as hybrid techniques. An extensive experimental evaluation shows that both algorithms generally improve over an exisiting heuristic from the literature.