Maximum Consistency Preferential Random Walks

Random walk plays a significant role in computer science. The popular PageRank algorithm uses random walk. Personalized random walks force random walk to “personalized views” of the graph according to users’ preferences. In this paper, we show the close relations between different preferential random walks and label propagation methods used in semi-supervised learning. We further present a maximum consistency algorithm on these preferential random walk/label propagation methods to ensure maximum consistency from labeled data to unlabeled data. Extensive experimental results on 9 datasets provide performance comparisons of different preferential random walks/label propagation methods. They also indicate that the proposed maximum consistency algorithm clearly improves the classification accuracy over existing methods.