ONE3A: one-against-all authentication model for smartphone using GAN network and optimization techniques.

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• Smartphones face computational limitations affecting authentication efficiency.
• Existing authentication methods may require additional hardware or incur high computational costs.
• Small datasets pose challenges for developing robust authentication models.

Purpose of the Study:

• To propose an efficient implicit authentication model for smartphones that overcomes computational constraints.
• To address the one-against-all classification problem in smartphone authentication.
• To enhance authentication accuracy using limited resources.

Main Methods:

• Integration of Conditional Tabular Generative Adversarial Network (CTGAN) for synthetic data generation to handle imbalanced datasets.
• Development of a novel feature selection technique based on the Whale Optimization Algorithm (WOA).
• Evaluation using the RHU touch mobile keystroke dataset, comparing WOA with Harris Hawks Optimization (HHO) and Random Forest (RF) classifier.

Main Results:

• The WOA-RF model achieved the highest classification accuracy (99.62% ± 0.40%) in mobile user authentication based on touch behavior.
• WOA demonstrated a superior reduction rate (87.85%) compared to the HHO algorithm.
• The proposed model effectively addresses the one-against-all classification challenge with improved efficiency.

Conclusions:

• The novel implicit authentication model integrating CTGAN and WOA-RF is highly effective for smartphone user authentication.
• This approach overcomes dataset limitations and computational constraints, offering a practical solution for mobile security.
• The WOA-RF model provides a significant advancement in enhancing authentication accuracy and efficiency on resource-limited devices.