Abstract:
The problem of learning an image classier that allows detection of out-of-distribution (OOD) examples, with the help of auxiliary background datasets, is studied. While training with background has been shown to improve OOD detection performance, the optimal choice of such dataset remains an open question, and challenges of data imbalance and computational complexity make it a potentially inefcient or even impractical solution. Targeted at balancing between efciency and detection quality, a dataset resampling approach is proposed for obtaining a compact yet representative set of background data points. The resampling algorithm takes inspiration from prior work on hard negative mining, performing an iterative adversarial weighting on the background examples and using the learned weights to obtain the subset of desired size. Experiments on different datasets, model architectures and training strategies validate the universal effectiveness and efciency of adversarially resampled background data. Code is available at https://github.com/JerryYLi/ bg-resample-ood.
