Timezone: »

Σ-Optimality for Active Learning on Gaussian Random Fields
Yifei Ma · Roman Garnett · Jeff Schneider

Thu Dec 05 07:00 PM -- 11:59 PM (PST) @ Harrah's Special Events Center, 2nd Floor

A common classifier for unlabeled nodes on undirected graphs uses label propagation from the labeled nodes, equivalent to the harmonic predictor on Gaussian random fields (GRFs). For active learning on GRFs, the commonly used V-optimality criterion queries nodes that reduce the L2 (regression) loss. V-optimality satisfies a submodularity property showing that greedy reduction produces a (1 − 1/e) globally optimal solution. However, L2 loss may not characterise the true nature of 0/1 loss in classification problems and thus may not be the best choice for active learning. We consider a new criterion we call Σ-optimality, which queries the node that minimizes the sum of the elements in the predictive covariance. Σ-optimality directly optimizes the risk of the surveying problem, which is to determine the proportion of nodes belonging to one class. In this paper we extend submodularity guarantees from V-optimality to Σ-optimality using properties specific to GRFs. We further show that GRFs satisfy the suppressor-free condition in addition to the conditional independence inherited from Markov random fields. We test Σ-optimality on real-world graphs with both synthetic and real data and show that it outperforms V-optimality and other related methods on classification.

Author Information

Yifei Ma (Amazon)

This is Yifei Ma. I work in Bayesian optimization, recommender systems, graph theory, and temporal modeling. Thanks.

Roman Garnett (Washington University in St. Louis)
Jeff Schneider (CMU)

More from the Same Authors