Please enable JavaScript.
Coggle requires JavaScript to display documents.
anomaly detection (technique used in detection (Proximity based relation…
anomaly detection
technique used in detection
Dependency based
:check:
relation among attributes
vertically
the role of variables
Dimension engineering
PCA-based method (hard PCA)
Mahalanobis Method (soft PCA)
Strenghth
: utlize inter-attribute correlation and extreme value analysis, which make it able to detect dependecy outlier and extreme value outlier simultaneously.
Replicator Neural network
Autoencoder
Anomaly Detection using Replicator Neural
Networks Trained on Examples of One Class
Outlier Detection using Replicator
Neural Networks
Orginal dimension
regression/classification model
ALSO
FRaC
LoPOD
time-series outlier detection
Bayesian Network :red_flag:
estimate the posterior probability/ joint probatility
(supervised, semi-supervised)
using local structure
making prediction
COMBN
feature selection
LoPOD
Neural Network
Proximity based
relation among objects
horizontally
distance-based
kNN method
density-based
LOF
angle-based
ABOD
Hybrid methods
Subspace-based
dependency is used to determine subspace :check:
HiCS
MBOM
(MB as subspace)
dependency is not involved in determining subspace
iForest
(random)
label avalability
Supervised
unsupervised
semisupervised
Need a model or not
model based
non-model based