Intelligent cyber security systems and policies of today rely on deep learning, a technology derived from artificial neural networks (ANNs). Cyber risk analytics that use artificial intelligence (AI) to boost business resilience and understand cyber security risk: pros and cons. Make use of deep learning tools like generative adversarial networks, deep belief networks, deep transfer learning, and deep reinforcement learning to intelligently combat cyber threats. There is also a set of tools that includes generic adversarial networks, auto-encoders, multilayer perceptrons, convolutional neural networks, recurrent neural networks (also known as long short-term memory), and so on. It is also possible to use these networks in ensembles or hybrid techniques. The goal of the backpropagation method is to convert inputs into outputs by making the most of the network weights. When training, a number of optimisation methods are employed, including Adam, Stochastic Gradient Descent, and Limited Memory BFGS. A wide range of cybersecurity issues could be addressed by these neural networks. Building reliable Internet of Things (IoT) systems, detecting harmful botnet traffic, analyzing malware, and conducting security threat analyses are all possible with MLP-based networks. Finding a computationally expensive solution to a complex security model may be difficult with MLP since the model is hyperparameter sensitive and needs adjustment of many different parameters, including the number of hidden layers, neurons, and iterations.