In this work, we present the first trial in the singlequery generalization of stochastic AUPRC optimization. For sharper generalization bounds, we focus on algorithm-dependent generalization.

Successful applications ofInfoNCE (Information Noise-ContrastiveEstimation) and its variants have popularized the use of contrastive variational mutual information (MI) estimators in machine learning.

Consideranews recommendation website that, when presented with a new user, sequentially offers a selection of currently trending articles. Such asystem may only haveafewopportunities tomakerecommendations before the user decides to navigate away, leaving little time to correct for misspecified or underspecified prior knowledge.

SAM aims to find a perturbed model within the vicinity of a current model that maximizes the loss over a training set.