Anomaly detection in web graphs using vertex neighbourhood based signature similarity methods

Imbalanced Aircraft Data Anomaly Detection

“…The mainstream flight-related anomaly detection methods are currently clustering-based [22]- [25], neighborhood-based [26]- [28], regression-based [29]- [31] and classificationbased [32]- [35] methods. Clustering-based [22]- [25] and neighborhood-based [26]- [28] methods require mining and exploiting relationships between data and require a high level of expert domain knowledge. Regression-based methods [29]- [31] detect anomalies by fitting to serial data and detecting anomalies ground on the residuals between inferred and actual values.…”

Positive and unlabeled learning for anomaly detection

“…There are plenty of works on anomaly detection problem using supervised and unsupervised methods. These techniques can be roughly divided into following categories: classification based methods [9][10][11][12], clustering based methods [13], nearest neighbor (NN) based methods [14], statistical [15], information theoretic [16] Many works consider anomaly problem for time-series datasets [18][19][20]. Jones et al [18] propose to transfer exemplar-based model for spotting anomalies in single dimension of the time series to multiple dimensions by a relation function.…”