Novelty Detection via Contrastive Learning with Negative Data Augmentation

TL;DR

This paper introduces a novel decoder-encoder framework for novelty detection that leverages contrastive learning and negative data augmentation, achieving state-of-the-art results with improved stability over GAN-based methods.

Contribution

The paper proposes a new decoder-encoder model with negative data augmentation and contrastive learning for more discriminative representations in novelty detection.

Findings

Outperforms existing novelty detection methods on benchmarks like CIFAR10 and DCASE.

Achieves state-of-the-art results with more stable training.

Utilizes a decoder-encoder framework with negative data augmentation for improved discriminative ability.

Abstract

Novelty detection is the process of determining whether a query example differs from the learned training distribution. Previous methods attempt to learn the representation of the normal samples via generative adversarial networks (GANs). However, they will suffer from instability training, mode dropping, and low discriminative ability. Recently, various pretext tasks (e.g. rotation prediction and clustering) have been proposed for self-supervised learning in novelty detection. However, the learned latent features are still low discriminative. We overcome such problems by introducing a novel decoder-encoder framework. Firstly, a generative network (a.k.a. decoder) learns the representation by mapping the initialized latent vector to an image. In particular, this vector is initialized by considering the entire distribution of training data to avoid the problem of mode-dropping. Secondly, a contrastive network (a.k.a. encoder) aims to ``learn to compare'' through mutual information estimation, which directly helps the generative network to obtain a more discriminative representation by using a negative data augmentation strategy. Extensive experiments show that our model has significant superiority over cutting-edge novelty detectors and achieves new state-of-the-art results on some novelty detection benchmarks, e.g. CIFAR10 and DCASE. Moreover, our model is more stable for training in a non-adversarial manner, compared to other adversarial based novelty detection methods.