Detecting Out-of-distribution Samples via Variational Auto-encoder with Reliable Uncertainty Estimation

TL;DR

This paper introduces INCPVAE, an improved variational autoencoder that enhances out-of-distribution detection and uncertainty estimation, addressing a key weakness of standard VAEs by better identifying OOD inputs and improving anomaly detection.

Contribution

The paper proposes INCPVAE, a novel scalable and trainable method that integrates an improved noise contrastive prior into VAEs for reliable OOD detection and uncertainty estimation.

Findings

INCPVAE outperforms standard VAEs in uncertainty estimation for OOD data.

The model demonstrates robustness in anomaly detection tasks.

Experiments confirm the effectiveness of INCPVAE across various datasets.

Abstract

Variational autoencoders (VAEs) are influential generative models with rich representation capabilities from the deep neural network architecture and Bayesian method. However, VAE models have a weakness that assign a higher likelihood to out-of-distribution (OOD) inputs than in-distribution (ID) inputs. To address this problem, a reliable uncertainty estimation is considered to be critical for in-depth understanding of OOD inputs. In this study, we propose an improved noise contrastive prior (INCP) to be able to integrate into the encoder of VAEs, called INCPVAE. INCP is scalable, trainable and compatible with VAEs, and it also adopts the merits from the INCP for uncertainty estimation. Experiments on various datasets demonstrate that compared to the standard VAEs, our model is superior in uncertainty estimation for the OOD data and is robust in anomaly detection tasks. The INCPVAE model obtains reliable uncertainty estimation for OOD inputs and solves the OOD problem in VAE models.