Deep Anomaly Detection with Outlier Exposure (ICLR 2019)
Automated Outlier Detection and Treatment Tool
Anomaly detection using LoOP: Local Outlier Probabilities, a local density based outlier detection method providing an outlier score in the range of [0,1].
[NeurIPS 2019 Spotlight] High dimensional mean estimation and outlier detection in nearly-linear time.
Anomaly detection related books, papers, videos, and toolboxes
BNP Paribas Kaggle Data Set Data source: https://www.kaggle.com/c/bnp-paribas-cardif-claims-management Outlier Detection- Ensemble unsupervised learning method - Isolation Forest The isolation algorithm is an unsupervised machine learning method used to detect abnormal anomalies in data such as outliers. This is once again a randomized & recursive partition of the training data in a tree structure. The number of sub samples and tree size is specified and tuned appropriately. The distance to the outlier is averaged calculating an anomaly detection score: 1 = outlier 0 = close to zero are normal data.
Code that implements the novel outlier detection algorithms from my Ph.D. dissertation.