Cyber security threats prediction is the security mechanism that detects malware or cyber-attacks in computer systems or the Internet of things. Cyber attacks change dynamically due to the mutations and modifications made by the attackers and lead to an ineffective prediction model. Most of the traditional learning models for cyber attack predictions learn the features to predict the future attack based on the stationary patterns, which is ineffective.
   Incremental learning is suitable to predict any form of cyberattacks. Incremental learning adaptively learns the new changes without forgetting the previously learned knowledge. The deep incremental learning model utilizes deep neural networks that automatically extract the complex patterns with long-range dependencies and high non-linearity to predict cyber threats. Deep incremental learning improves the effectiveness and efficiency of the cyber security threats prediction model.