Journal
IEEE ACCESS
Volume 9, Issue -, Pages 43991-44002Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/ACCESS.2021.3065838
Keywords
Anomaly detection; Training; Image reconstruction; Image restoration; Deep learning; Data models; Task analysis; Deep Learning; unsupervised outlier detection; autoencoder; transformation invariant autoencoder
Categories
Funding
- National Key Research and Development Program of China [2018YFB0204301]
- National Natural Science Foundation of China [62006236]
- Hunan Provincial Natural Science Foundation [2020JJ5673]
- National University of Defense Technology (NUDT) Research Project [ZK20-10]
Ask authors/readers for more resources
The paper introduces a framework called Transformation Invariant AutoEncoder (TIAE) that achieves stable and high performance on unsupervised outlier detection for complex image datasets. By improving the autoencoder and incorporating adaptive self-paced learning, TIAE significantly advances the performance of unsupervised outlier detection by up to 10% AUROC compared to other autoencoder based methods on five image datasets.
Autoencoder based methods are the majority of deep unsupervised outlier detection methods. However, these methods perform not well on complex image datasets and suffer from the noise introduced by outliers, especially when the outlier ratio is high. In this paper, we propose a framework named Transformation Invariant AutoEncoder (TIAE), which can achieve stable and high performance on unsupervised outlier detection. First, instead of using a conventional autoencoder, we propose a transformation invariant autoencoder to do better representation learning for complex image datasets. Next, to mitigate the negative effect of noise introduced by outliers and stabilize the network training, we select the most confident inliers likely examples in each epoch as the training set by incorporating adaptive self-paced learning in our TIAE framework. Extensive evaluations show that TIAE significantly advances unsupervised outlier detection performance by up to 10% AUROC against other autoencoder based methods on five image datasets.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available