<ul data-eligibleForWebStory="true"><li>Unsupervised anomaly detection is a challenging task utilizing autoencoders and generative models.</li><li>Autoencoders are often used to model normal data distribution and identify anomalies by high reconstruction error.</li><li>The proposed approach involves a decoupled training using both an autoencoder and a likelihood model with noise contrastive estimation (NCE).</li><li>NCE estimates a probability density function for anomaly scoring in the joint space of the autoencoder's latent representation and reconstruction quality features.</li><li>To improve NCE's false negative rate, reconstruction features are systematically varied during training to optimize the noise distribution.</li><li>Experimental assessments on multiple benchmark datasets show that the proposed approach matches the performance of leading anomaly detection algorithms.</li></ul>

Anomaly Detection via Autoencoder Composite Features and NCE

Discover more