Isolation Forest

By Fei Tony Liu, Kai Ming Ting, Zhi-Hua Zhou. 2008

6,516 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Control and Systems Engineering, Statistics and Probability, Anomaly & Outlier Detection

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Control and Systems Engineering

In Statistics and Probability

In Anomaly & Outlier Detection

Citations in the corpus

BoostClean: Automated Error Detection and Repair for Machine Learning (2017, methodology): “We cast the former component into a featurization problem so that the user focuses on the familiar task of creating feature extraction functions while BoostClean translates these features into error detection rules using a technique called Isolation Forests [35].”
Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shift (2018, methodology): “Popular approaches to anomaly detection include density estimation [6], margin-based approaches such as the one-class SVM [40], and the tree-based isolation forest method due to [30].”
Learning Placeholders for Open-Set Recognition (2021, background): “The problem of out-ofdistribution detection [10, 17, 1], anomaly detection [4, 18, 25] and novelty detection [11, 32, 41, 21, 43] are related topics to open-set recognition.”
Open Category Detection with PAC Guarantees (2018, methodology): “We carry out experiments1 on synthetic and benchmark datasets using a state-of-the-art anomaly detector, the Isolation Forest (Liu et al., 2008).”
You Are What You Do: Hunting Stealthy Malware via Data Provenance Analysis (2020, methodology): “The three baseline methods are as follows: • Isolation Forest [72]: This algorithm divides the data points to different partitions.”
Deep Semi-Supervised Anomaly Detection (2019, background): “Shallow anomaly detectors such as the One-Class SVM (OC-SVM) [60], Support Vector Data Description (SVDD) [65], Isolation Forest (IF) [38], or Kernel Density Estimator (KDE) [50, 31, 68] often require manual feature engineering to be effective on high-dimensional data and are limited in their scalab”
PyOD: A Python Toolbox for Scalable Outlier Detection (2019, background): “, 2008) Proximity Yes No Isolation Forest (Liu et al., 2008) Ensembling No Yes HBOS (Goldstein and Dengel, 2012) Proximity Yes No SOS (Janssens et al.”
Progress in Outlier Detection Techniques: A Survey (2019, methodology): “[181], several additional methods have been proposed [38], [182], [190]–[192] to improve outlier detection using ensemble methods, but most of these methods are meta methods except for those suggested by [37].”
Adversarially Learned Anomaly Detection (2018, background): “Isolation Forests (IF) [35] are a newer classic machine learning technique that isolates anomalies instead of modeling the distribution of normal data.”
Unknown Intent Detection Using Gaussian Mixture Model with an Application to Zero-shot Intent Classification (2020, methodology): “…the open-world classiﬁcation as an outlier detection problem by exploiting anomaly detection methods such as robust covariance estimators (Rousseeuw and Driessen, 1999), one-class SVM (Sch¨olkopf et al., 2001), isolation forest (Liu et al., 2008) and local outlier factor (Breunig et al., 2000).”
Learning Bounds for Open-Set Learning (2021, methodology): “Thus, we recommend using the outlier sample score (with range [0 , 1] ) given by isolation forest (iForest) (Liu et al., 2008) as the sample weights, which has time complexity O (( n + m ) log( n + m )) .”
Sequence classification for credit-card fraud detection (2018, background): “Anomaly scores can be derived directly from the likelihood in probabilistic models, from the average path lengths in ensembles of decision trees [Liu et al., 2008], the local density in the neighbourhood around an instance [Breunig et al.”
OpenOOD: Benchmarking Generalized Out-of-Distribution Detection (2022, methodology): “Similarly, AD methods can apply to OOD detection task by ignoring all the labels within the in-distribution [32, 33].”
Exploring Backdoor Poisoning Attacks Against Malware Classifiers (2020, methodology): “First, the Isolation Forest, trained on the reduced feature space, is often capable of correctly isolating all the backdoored points with relatively low false positives.”
ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions (2022, methodology): “Notable works of ensemble-based OD include feature bagging [25] that use various sub-feature spaces, isolation forests [17] that aggregate the information from multiple base trees, LSCP [18] that dynamically picks the best base estimator for each data point, and SUOD [21] that uses a large number…”
Classification-Reconstruction Learning for Open-Set Recognition (2018, methodology): “Some of the generic methods for anomaly detection are oneclass extension of discriminative models such as SVM [24] or forests [21], generative models such as Gaussian mixture models [33], and subspace methods [32].”
Deep Learning for Unsupervised Insider Threat Detection in Structured Cybersecurity Data Streams (2017, methodology): “However, it is common in anomaly detection literature (Malhotra et al. 2016) to use an auto-encoder to detect anomaly.”
HRN: A Holistic Approach to One Class Learning (2020, background): “…one-class classiﬁcation based methods [75, 78, 39], deep learning based representation learning using auto-encoders [10, 89, 69, 7, 94] and adversarial learning [74, 15, 64, 20], ensemble methods [53, 10], graphs and random walks [58, 31], transfer learning [45, 2], and multi-task learning [35].”