Anomaly detection is a challenging task for machine learning algorithms . DA3D uses adversarial autoencoders to generate counterexamples based on the normal data only . These artificialanomalies used during training allow the detection of real, yet unseen, real, unseen anomalies . With our novel generative approach, we transform the unsupervised task of anomaly detection to a supervised one, which is more tractable by machine learning and especially deep learning methods, such as deep learning, it is more difficult to find out if there are any known anomalies if any are available to manually analyze data, thus usually only few known anomalies are available .

Author(s) : J. -P. Schulze, P. Sperl, K. Böttinger

Links : PDF - Abstract

Code :

https://github.com/santi-pdp/pase


Coursera

Keywords : learning - detection - anomaly - anomalies - adversarial -

Leave a Reply

Your email address will not be published. Required fields are marked *