Towards Maximizing the Representation Gap between In-Domain & Out-of-Distrib

Towards Maximizing the Representation Gap between In-Domain & Out-of-Distribution Examples

Among existing uncertainty estimation approaches, Dirichlet Prior Network (DPN) distinctly models different predictive uncertainty types. However, for in-domain examples with high data uncertainties among multiple classes, even a DPN model often produces indistinguishable representations from the out-of-distribution (OOD) examples, compromising their OOD detection performance.We address this shortcoming by proposing a novel loss function for DPN to maximize the \textit{representation gap} between in-domain and OOD examples. Experimental results demonstrate that our proposed approach consistently improves OOD detection performance.

努力使域内和分布外示例之间的表示差距最大化

在现有的不确定性估计方法中，狄利克雷特先验网络（DPN）可以对不同的预测不确定性类型进行建模。但是，对于在多个类别之间具有较高数据不确定性的域内示例，即使DPN模型也经常会与分布外（OOD）示例产生无法区分的表示，从而损害了其OOD检测性能。.. 我们通过提出一种新颖的损失函数解决这一缺点对于DPN最大化\ textit {表示间隙}域内和OOD示例之间。实验结果表明，我们提出的方法可以持续提高OOD检测性能。 （阅读更多）