GradStop uses gradient magnitude sampling and cohesion metrics to stop unsupervised deep outlier detection training when inlier priority weakens, improving AUC.
Hyperparameter sensitivity in deep outlier detection analysis and a scalable hyper-ensemble solution
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
GradStop: Exploring Training Dynamics in Unsupervised Outlier Detection through Gradient
GradStop uses gradient magnitude sampling and cohesion metrics to stop unsupervised deep outlier detection training when inlier priority weakens, improving AUC.