A Hybrid Model for Anomaly Detection in Cloud Using Deep Learning
Dhanujakshi1, Dr. Ravinarayana B2,
1Master’s in Technology Student, Dept. of Computer Science and Engineering, Mangalore institute of technology and Engineering, Moodbidri,
2HOD & Associate Professor, Dept. of Computer Science and Engineering, Mangalore institute of technology and Engineering, Moodbidri,
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The exponential growth of cloud infrastructure and services has resulted in a sharp increase in the demand for effective anomaly detection methods. Traditional rule-based and statistical methodologies have a tough time keeping up with cloud systems because of their dynamic and complex nature. As a result, there is a growing need to apply deep learning techniques to address the issue. Here we provide a unique method for utilizing deep learning to identify anomalies in cloud environments. One of the main contributions of this research is the design and implementation of deep learning-based anomaly detection system that is specially tailored to the cloud’s data properties. Convolution Neural Networks and auto encoders are used to extract the patterns and spatial representations from structured cloud data. Thorough experiments are conducted on a large scale cloud dataset with a range of usage scenarios and anomaly types to evaluate the effectiveness of the recommended technique. The deep learning model has benefits over more conventional anomaly detection methods when compared in terms of accuracy, sensitivity and false-positive rates, according to analyses. The findings of my analysis suggest that deep learning based LSTM and RSU algorithms significantly increase the security and reliability of network data by identifying abnormal behaviors. The RSU algorithm forecasts an accuracy of 99 comparable to the LSTM algorithm.
Key Words: Convolution Neural Networks, LSTM algorithm, RSU algorithm, Deep Learning
