Anomaly Detection Using Maximum Entropy Fuzzy Clustering Algorithm Enhanced with Soft Computing Techniques

With the continuous growth of data volume, anomaly detection has become an important link in the data processing process. In view of the maximum entropy fuzzy clustering algorithm, an anomaly detection method combining soft computing is proposed. During the process, the K-means algorithm was used to...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Informatica (Ljubljana) Ročník 48; číslo 17; s. 171 - 182
Hlavný autor: Liang, Chunhua
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Ljubljana Slovenian Society Informatika / Slovensko drustvo Informatika 01.11.2024
Predmet:
Accuracy
Algorithms
Anomalies
Cluster analysis
Clustering
Data processing
Datasets
Efficiency
Entropy
Experiments
Information technology
Machine learning
Maximum entropy
Neural networks
Purity
Research methodology
Soft computing
ISSN:0350-5596, 1854-3871
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:With the continuous growth of data volume, anomaly detection has become an important link in the data processing process. In view of the maximum entropy fuzzy clustering algorithm, an anomaly detection method combining soft computing is proposed. During the process, the K-means algorithm was used to construct the algorithm foundation, followed by the establishment of an objective function for maximum entropy calculation and the introduction of the Hilbert Schmidt independence criterion for variable extraction. Then it conducts data migration and calculates the exception score. The experimental results showed that the proposed method could be reduced to 113 in the Iris data set when the convergence curve was tested. When the calculation time was tested, the calculation time of the research method was only 2697ms when the sample size reached 10000. When the accuracy and purity tests were carried out, the accuracy and purity of the research method were 87.7% and 87.6% in the MR Dataset. In the Leaf dataset, the standardized mutual information index reached 0.6837 and the FM index reached 0.3903. The lowest Davies-Bouldin index was 0.71. The area enclosed by the receiver operation characteristic curve and the horizontal coordinate of the research method was the largest. The results indicate that the research method has high accuracy and computational efficiency in data anomaly detection and can provide effective technical references for anomaly detection.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0350-5596
1854-3871
DOI:10.31449/inf.v48i17.6537