Papers › Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection

Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection

1 Jan 2018ICLR 2018 1archive 2025-07-28

Bo Zong, Qi Song, Martin Renqiang Min, Wei Cheng, Cristian Lumezanu, Daeki Cho, Haifeng Chen

Unsupervised anomaly detection on multi- or high-dimensional data is of great importance in both fundamental machine learning research and industrial applications, for which density estimation lies at the core. Although previous approaches based on dimensionality reduction followed by density estimation have made fruitful progress, they mainly suffer from decoupled model learning with inconsistent optimization goals and incapability of preserving essential information in the low-dimensional space. In this paper, we present a Deep Autoencoding Gaussian Mixture Model (DAGMM) for unsupervised anomaly detection. Our model utilizes a deep autoencoder to generate a low-dimensional representation and reconstruction error for each input data point, which is further fed into a Gaussian Mixture Model (GMM). Instead of using decoupled two-stage training and the standard Expectation-Maximization (EM) algorithm, DAGMM jointly optimizes the parameters of the deep autoencoder and the mixture model simultaneously in an end-to-end fashion, leveraging a separate estimation network to facilitate the parameter learning of the mixture model. The joint optimization, which well balances autoencoding reconstruction, density estimation of latent representation, and regularization, helps the autoencoder escape from less attractive local optima and further reduce reconstruction errors, avoiding the need of pre-training. Experimental results on several public benchmark datasets show that, DAGMM significantly outperforms state-of-the-art anomaly detection techniques, and achieves up to 14% improvement based on the standard F1 score.

PaperPDFCode

Code

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Anomaly DetectionDensity EstimationDimensionality ReductionUnsupervised Anomaly DetectionUnsupervised Anomaly Detection with Specified Settings -- 0.1% anomalyUnsupervised Anomaly Detection with Specified Settings -- 1% anomalyUnsupervised Anomaly Detection with Specified Settings -- 10% anomalyUnsupervised Anomaly Detection with Specified Settings -- 20% anomalyUnsupervised Anomaly Detection with Specified Settings -- 30% anomaly

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Unsupervised Anomaly Detection with Specified Settings -- 0.1% anomaly CIFAR-10 DAGMM AUC-ROC 0.494 #6 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 0.1% anomaly Cats and Dogs DAGMM AUC-ROC 0.784 #4 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 0.1% anomaly Fashion-MNIST DAGMM AUC-ROC 0.784 #4 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 0.1% anomaly MNIST DAGMM AUC-ROC 0.624 #4 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 0.1% anomaly STL-10 DAGMM AUC-ROC 0.574 #5 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 1% anomaly CIFAR-10 DAGMM AUC-ROC 0.503 #6 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 1% anomaly Cats and Dogs DAGMM AUC-ROC 0.710 #5 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 1% anomaly Fashion-MNIST DAGMM AUC-ROC 0.793 #5 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 1% anomaly MNIST DAGMM AUC-ROC 0.708 #5 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 1% anomaly STL-10 DAGMM AUC-ROC 0.477 #6 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 10% anomaly CIFAR-10 DAGMM AUC-ROC 0.778 #6 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 10% anomaly Cats and Dogs DAGMM AUC-ROC 0.960 #3 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 10% anomaly Fashion-MNIST DAGMM AUC-ROC 0.788 #4 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 10% anomaly MNIST DAGMM AUC-ROC 0.629 #5 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 10% anomaly STL-10 DAGMM AUC-ROC 0.826 #5 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 20% anomaly Cats and Dogs DAGMM AUC-ROC 0.914 #3 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 20% anomaly Fashion-MNIST DAGMM AUC-ROC 0.780 #4 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 20% anomaly MNIST DAGMM AUC-ROC 0.616 #5 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 20% anomaly STL-10 DAGMM AUC-ROC 0.911 #4 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 20% anomaly cifar10 DAGMM AUC-ROC 0.883 #2 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 30% anomaly ASSIRA Cat Vs Dog DAGMM AUC-ROC 0.846 #1 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 30% anomaly CIFAR-10 DAGMM AUC-ROC 0.850 #2 of 6 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 30% anomaly Fashion-MNIST DAGMM AUC-ROC 0.769 #4 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 30% anomaly MNIST DAGMM AUC-ROC 0.613 #5 of 5 Archive leaderboard report
Unsupervised Anomaly Detection with Specified Settings -- 30% anomaly STL-10 DAGMM AUC-ROC 0.883 #4 of 6 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections