Active Learning on Attributed Graphs via Graph Cognizant Logistic Regression and Preemptive Query Generation

Node classification in attributed graphs is an important task in multiple\npractical settings, but it can often be difficult or expensive to obtain\nlabels. Active learning can improve the achieved classification performance for\na given budget on the number of queried labels. The best existing methods are\nbased on graph neural networks, but they often perform poorly unless a sizeable\nvalidation set of labelled nodes is available in order to choose good\nhyperparameters. We propose a novel graph-based active learning algorithm for\nthe task of node classification in attributed graphs; our algorithm uses graph\ncognizant logistic regression, equivalent to a linearized graph convolutional\nneural network (GCN), for the prediction phase and maximizes the expected error\nreduction in the query phase. To reduce the delay experienced by a labeller\ninteracting with the system, we derive a preemptive querying system that\ncalculates a new query during the labelling process, and to address the setting\nwhere learning starts with almost no labelled data, we also develop a hybrid\nalgorithm that performs adaptive model averaging of label propagation and\nlinearized GCN inference. We conduct experiments on five public benchmark\ndatasets, demonstrating a significant improvement over state-of-the-art\napproaches and illustrate the practical value of the method by applying it to a\nprivate microwave link network dataset.\n