REVIEW 4 cited by
How to Evaluate the Quality of Unsupervised Anomaly Detection Algorithms?
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
When sufficient labeled data are available, classical criteria based on Receiver Operating Characteristic (ROC) or Precision-Recall (PR) curves can be used to compare the performance of un-supervised anomaly detection algorithms. However , in many situations, few or no data are labeled. This calls for alternative criteria one can compute on non-labeled data. In this paper, two criteria that do not require labels are empirically shown to discriminate accurately (w.r.t. ROC or PR based criteria) between algorithms. These criteria are based on existing Excess-Mass (EM) and Mass-Volume (MV) curves, which generally cannot be well estimated in large dimension. A methodology based on feature sub-sampling and aggregating is also described and tested, extending the use of these criteria to high-dimensional datasets and solving major drawbacks inherent to standard EM and MV curves.
Forward citations
Cited by 4 Pith papers
-
Analyzing Shapley Additive Explanations to Understand Anomaly Detection Algorithm Behaviors and Their Complementarity
SHAP attribution profiles can identify complementary anomaly detectors whose divergence in explanations predicts non-overlapping detections, enabling stronger ensembles when high individual performance is maintained.
-
Automatic Unsupervised Ensemble Outlier Model Selection--Extended Version
MetaEns trains on meta-datasets to predict marginal gains from adding models and uses a submodular-inspired objective with diversity discounting and risk regularization for greedy unsupervised ensemble selection, outp...
-
Self-Adaptive Anomaly Detection with Reinforcement Learning and Human Feedback in Connected Vehicles
Attention-augmented factorized DQN detector selection plus conjunctive drift alarms and 60/40 replay recovers F1 from 0.52 to 0.65 after real concept drift without catastrophic forgetting on a seven-service valet-park...
-
Leveraging the Christoffel Function for Outlier Detection in Data Streams
Two Christoffel-function methods, DyCF and DyCG, detect outliers in low-dimensional data streams with one hyperparameter or none.
Discussion (0). Continue with ORCID to comment.