Principal Component-based Anomaly Detection Scheme

MANOVA- A Multivariate Statistical Variance Analysis for WSN Using PCA

“…The authors determine that a PCA global scattered modelling attains higher discovery presentation than the centralized modelling for sinkhole assaults (sinkhole occurrences are those wherein a malicious node sends fake routing information claiming a most desirable course to make different nodes course statistics packets thru the malicious node to inspect and filter the visitors). A PCA-based totally anomaly detection is proposed in [15]. In that reference, the authors broaden a gadget with phases: information modelling and anomaly detection.…”

UNPCC: A Novel Unsupervised Classification Scheme for Network Intrusion Detection

“…Principal Component Analysis (PCA), the core of the PCC (Principal Component Classifier) anomaly detection algorithm that we developed [10] [11], is employed to reduce the dimension of a training data set, allowing for further data analysis and easier exploration of the statistical information present in the data set. Principal components are particular linear combinations of the original variables with two important properties:…”

“…Apply PCA to the whole training data set to obtain the principal components, their eigenvalues, and data instance scores as was done in [10] [11]. Define Y= y ij , i = 1, 2, .…”

Exploração das propriedades de hubness para detecção semissupervisionada de outliers em dados de alta dimensão

“…Após essa redução nos dados, é realizado a normalização dos dados a partir da ponderação entre a média e o desvio padrão dos dados. Depois, o algoritmo PCA (SHYU et al, 2003) realiza a análise da significância de cada variável. O algoritmo PCA resultará no autovalor e autovetor dos dados normais.…”