Anomaly Based Security and Preventive Measures in Big Data Using Artificial Neural Network (ANN)

In this research article the researcher proposed anomaly-based security systems identify deviations from established normal behavior patterns to detect potential threats. The proposed systems used the machine learning to learn what constitutes normal activity and flag any unusual behavior as a potential anomaly. This proactive approach helps in detecting unknown or zero-day attacks that signature-based systems might miss. The research used the Artificial Neural Networks (ANNs) which are powerful tool for analyzing big data, offering several advantages over traditional statistical methods. ANNs can learn complex patterns from large datasets, handle incomplete or noisy data, and make predictions with high speed and accuracy. To train the ANN machine learning algorithms the researcher used the 80% of 125972 datasets with 43 features and test the model accuracy the researcher used 20% of 125972 datasets. The researcher found that model training accuracy 0.95% and test accuracy: 0.96% that does deal with the problem of fast machine learning on large datasets, it accomplishes so in a manner that is independent of applications and implementation for improving the security system .