Randomized shortest paths and their applications
Silvia García Díez · 2015
In graph analysis, the Shortest Path problem identifies the optimal, most cost effective, path between two nodes. This problem has been the object of many studies and extensions in heterogeneous domains such as: speech recognition, social network analysis, biological sequence alignment, path planning, or zero-sum games among others. Although the shortest path focuses on the optimal cost of reaching a destination node, it does not take into account other useful information contained on the graph, such as the degree of connectivity of two nodes. On the other hand, measures taking connectivity information into account have their own drawbacks, specially when graphs become large. A new family of distances which interpolates between both extremes is introduced by the Randomized Shortest Path (RSP) framework. By spreading randomization through a graph, the RSP leads to applications where some degree of randomness would be desired. Through this work, we try to investigate whether the RSP framework can be applied to different domains in which randomization is useful, and either solve an existing problem with a new approach, or prove to outperform existing methods.