AI-ordlista
Den kompletta ordlistan över AI
Local density-based detection (LOF)
Method based on comparing the local density of a point with that of its neighbors to identify outliers.
Isolation Forest
Ensemble algorithm that isolates observations by building random decision trees to detect anomalies.
Autoencoders for anomalies
Neural networks that learn to reconstruct normal data and identify anomalies by high reconstruction error.
One-Class SVM
Support vector machine that learns a decision boundary around normal data to detect outliers.
Time series anomaly detection
Specialized techniques for identifying unusual patterns in temporal sequential data.
Multivariate anomaly detection
Identification of anomalous observations based on complex relationships between multiple variables simultaneously.
Detection by clustering (DBSCAN)
Using clustering algorithms where points not belonging to any cluster are considered as anomalies.
Data stream detection
Real-time methods to identify anomalies in continuously arriving data without complete storage.
GANs for anomaly detection
Generative Adversarial Networks used to model the normal distribution and detect unlikely samples.
Graph anomaly detection
Identification of unusual nodes, edges or subgraphs in relational data structures
Contextual anomaly detection
Detection of abnormal observations only in a specific context, based on environmental conditions
Collective anomaly detection
Identification of groups of observations that are collectively abnormal even if individually normal.
Robust statistical methods
Approaches based on outlier-resistant statistics such as medians or robust quantiles.
High-dimensional anomaly detection
Specialized techniques to handle the curse of dimensionality in multivariate outlier detection.
Semi-supervised learning for anomalies
Approaches combining labeled and unlabeled data to improve anomaly detection with few examples.