User Authentication in mobile sensors using RNN and SOA Algorithm

Implicit verification systems prevent security and threat risks that demonstrate behavior for cell phones. Nevertheless, of late, experts have revealed that the presentation of social biometrics is not enough. Hence, in this paper, develop a hybrid approach for gait-based user authentication using mobile sensors. The Recurrent Neural Network (RNN) and Seagull Optimization Algorithm (SOA) are combined in the suggested hybrid approach. The best network parameters will be chosen using the SOA in order to enhance RNN performance. The principle purpose of the work is to identify a pair of class customer with a character tricky where a person who is a real customer of a cell phone is labeled authenticated, while any remaining person is designated as non-authenticated. There are four stages involved in the suggested methodology: pre-processing, feature extraction, activity recognition, and authentication. An average smoothing filter helps to remove the unwanted noise during the pre-processing stage. Twelve distinct features are extracted during the feature extraction phase. The suggested method is applied in the activity recognition stage in order to identify the activities. The probabilistic scoring model is utilized for user validation during the user authentication phase. Finally, the user validation is identified based on their activities which is executed and validated. To evaluate the performance of the proposed methodology, it is associated with the traditional techniques such as RNN-Particle Swarm Optimization (PSO) and RNN- (Genetic Algorithm).