Particle Swarm Optimization-Tuned Deep Neural Network for Cloud Intrusion Detection

Cloud computing has transformed the contemporary digital platforms because it provides specifiable, flexible, and affordable services. Nonetheless, its dynamism and multi-tenant have seen it become a first-choice target of most types of cyberattacks. There is often no current security mechanism that is automated which helps in detecting complicated intrusion attempts and hence there is a need to incorporate smart and adaptive security. In this paper, we think of a new particle-swarm-optimization-tuned deep-neural-network (PSO-DNN) framework to be used to successfully detect intrusion in the cloud network. The model takes advantage of the central learning power of DNNs to identify complicated patterns and applies PSO to optimize the most important hyperparameters including learning rates, the number of hidden layers and the number of neurons. The suggested method is tested on standard intrusion detection data sets and it shows a great increase in accuracy of classification, the rate of detection and the minimization of false positives when compared to plain conventional DNN models that are not optimized. Based on experiments, PSO improves the quickness of convergence and model resilience, which makes it exceptionally useful to deal with enormous scale, high-dimensional traffic data in a cloud. The PSO-DNN model proposed presents a scalable and efficient solution to real-time threats detection hence enhancing cloud security to known attacks and zero-day attacks too. The given paper provides an understanding of how hybrid intelligent systems can future proof cloud systems.