Deep Learning-Based Authentication Using Keystroke-Dynamics

Keystroke dynamics where a user is authenticated based on his/her typing patterns. It is considered as best solution to authenticate a user as this problem is solved by considering behavioural characteristics which is very difficult to copy. In this research paper we solved the problem of static keystroke dynamics by deep learning approach.in this paper we use the concept of quantile transformation which reduces the impact of outliers. For pattern reorganization, various optimization algorithms are used. For global pattern recognition various Metaheuristics algorithms and for local pattern ADAM optimization algorithm is used. The best solution is achieved by the Firefly Optimization Algorithm (a nature-inspired, swarm-based metaheuristic) which excels at global pattern reorganization by using bioluminescent-based attraction. Here, less-optimal solutions are drawn toward better ones through intensity-based movements, with attractiveness decreasing over distance. This mechanism enables efficient exploration of the search space and helps locate the best solution.