AI-ordlista
Den kompletta ordlistan över AI
Density-based anomaly detection
Approach that identifies anomalies as points located in low-density regions compared to the rest of the data, using algorithms such as LOF or DBSCAN.
Real-time anomaly detection
Continuous process of identifying anomalies in instantaneous data streams, requiring low-latency and high-performance algorithms.
Multivariate anomalies
Abnormal deviations detected when considering multiple variables simultaneously, which may go unnoticed in univariate analysis.
Contextual anomalies
Observations that are abnormal only in a specific context, such as a high sale during normal periods but low during sales.
Collective anomalies
Set of observations that are normal individually but abnormal when they appear together in a sequence or group.
Grubbs' test
Statistical hypothesis test for detecting a single outlier in a normally distributed dataset.
Deep learning-based anomaly detection
Use of deep neural networks such as autoencoders, GANs, or LSTMs to model complex patterns and identify deviations.
Supervised anomaly detection
Approach using labeled data (normal/anomalous) to train classification models such as logistic regression or random forests.
Métrique de reconstruction
Erreur quadratique moyenne ou autre mesure de divergence entre les données originales et leur reconstruction par un modèle, utilisée pour quantifier l'anormalité.