Dissimilar Nodes Improve Graph Active Learning

Training labels for graph embedding algorithms could be costly to obtain in many practical scenarios. Information score-based active learning (AL) algorithms are frameworks that could obtain the most useful labels for training while keeping the total label queries under a certain budget. The existing Active Graph Embedding framework has been shown to be capable of bringing some improvement to the node classification tasks of Graph Convolutional Networks, however, when evaluating the importance of unlabeled nodes, they fail to consider the influence of existing labeled nodes on the value of unlabeled nodes. With the aim of addressing this limitation, in this work, we introduce 3 dissimilarity-based information scores for active learning: feature dissimilarity score (FDS), structure dissimilarity score (SDS), and embedding dissimilarity score (EDS). According to experiments, our newly proposed scores boost the classification accuracy by 2.1$\%$ on average and are capable of generalizing to different Graph Neural Network architectures.