OPHiForest: Order Preserving Hashing Based Isolation Forest for Robust and Scalable Anomaly Detection

Control Charts and Machine Learning for Anomaly Detection in Manufacturing

“…Finally, we avoided to fill missing outcomes for such methods, like GIF, where the authors provided a fine-tuning of their algorithms, since our last assumptions would not replicate the same performances. [98] 1.00 [98] 0.59 [11] 0.71 [98] 0.65 [98] 0.98 [98] 0.72 [98] 0.91 [93] EIF 0.99 0.85 [37] 0.92 0.99 [37] 0.86 0.78 0.70 0.99 0.80 [37] 0.91 RRCF 0.99 [29] 0.89 [29] 0.91 0.91 [18] 0.83 [18] 0.68 [18] 0.59 [18] 0.64 [18] 0.74 [18] 0.90 [18] IMF [18] 0.99 -0.84 [18] OPHiForest [53], Split-Criteria Isolation Forest (SCIForest) [54], Extended Isolation Forest (EIF) [33], Robust Random Cut Forest (RRCF) [30], Isolation Mondrian Forest (IMF) [60], Mondrian Polya Forest (MPF) [18], Partial Identification Forest (PIDForest) [29], Order Preserving Hashing Based Isolation Forest (OPHIF) [93], Locality Sensitive Hashing Isolation Forest (LSHiForest) [98], Hybrid Isolation Forest (HIF) [64], Hybrid Extended Isolation Forest (HEIF) [37], One-class Random Forest (OneClassRF) [27], Trident Forest (T-Forest) [97], Entropy-based Greedy Isolation Tree (EGiTree) [50], Generalized Isolation Forest (GIF) [11], Distribution Forest (dForest) [95], Re-Mass Isolation Forest (ReMass IF) [6], Half-Spaces Forest (HSF) [84]. Scores are referenced when are provided by a different source than the original paper for the method.…”

“…The work shown in [93] improves on the core ideas of the LSHiForest by proposing a learning to hash (LTH) method to select the hashing function which best preserve similarities in the dataset in the projected space. The order preserving hashing algorithm (OPH) is chosen for such task as it shows excellent performances in nearest neighbour search.…”

Detecting anomalies within smart buildings using do-it-yourself internet of things

“…Isolation Forest (IF) was used to detect anomalies in smart audio sensors (Antonini et al 2018). IF is also used, in combination with order-preserving hashing techniques, to detect anomalies by Xiang et al (2020). Another novel approach proposed by Farzad and Gulliver T (2020) uses autoencoder based IF for log-based anomaly detection.…”