VOS: Learning What You Don't Know by Virtual Outlier Synthesis
This addresses the challenge of OOD detection for safe deployment of neural networks, offering a novel approach that avoids reliance on real outlier datasets, though it is incremental in improving existing methods.
The paper tackles the problem of out-of-distribution (OOD) detection in neural networks by proposing VOS, a framework that synthesizes virtual outliers for model regularization, achieving a reduction in FPR95 by up to 9.36% compared to previous methods.
Out-of-distribution (OOD) detection has received much attention lately due to its importance in the safe deployment of neural networks. One of the key challenges is that models lack supervision signals from unknown data, and as a result, can produce overconfident predictions on OOD data. Previous approaches rely on real outlier datasets for model regularization, which can be costly and sometimes infeasible to obtain in practice. In this paper, we present VOS, a novel framework for OOD detection by adaptively synthesizing virtual outliers that can meaningfully regularize the model's decision boundary during training. Specifically, VOS samples virtual outliers from the low-likelihood region of the class-conditional distribution estimated in the feature space. Alongside, we introduce a novel unknown-aware training objective, which contrastively shapes the uncertainty space between the ID data and synthesized outlier data. VOS achieves competitive performance on both object detection and image classification models, reducing the FPR95 by up to 9.36% compared to the previous best method on object detectors. Code is available at https://github.com/deeplearning-wisc/vos.