Out-of-Distribution Detection with An Adaptive Likelihood Ratio on Informative Hierarchical VAE

Unsupervised out-of-distribution (OOD) detection is essential for the reliability of machine learning. In the literature, existing work has shown that higher-level semantics captured by hierarchical VAEs can be used to detect OOD instances. However, we empirically show that, the inheirt “ posterior collapse ” of hierarchical VAEs would seriously limit their capacity for OOD detection. Based on a thorough analysis, we propose an informative hierarchical VAE to alleviate this issue through enhancing the connections between the data sample and its multi-layer stochastic latent representations during training. Furthermore, we propose a novel score function for unsupervised OOD detection, referred to as Adaptive Likelihood Ratio, which can selectively aggregate the semantic information on multiple hidden layers of hierarchical VAEs, leading to a strong separability between in-distribution and OOD samples. Experimental results demonstrate that our method can significantly outperform existing state-of-the-art unsupervised OOD detection approaches.