Designing Secure and Usable Systems: The Intersection of Human-Computer Interaction, Cybersecurity, and Machine Learning

Le résumé fourni par la source

In this fast-paced world of IT security, HCI, cybersecurity, and machine learning initiatives should lead to design that is strong and easy to use technological systems.This study assesses the capabilities of three notable ML models-ANN, CNN, and SVM-in making cybersecurity better, utilizing the UNSW-NB15 data set.Using an 80-20 train-test split and 5-fold crossvalidation of the data, the CNN model showed to be the best across the three models.This is because it generated an accuracy of 95.3% with a precision of 94.5% and recall of 96.0%, among others.All in all, the CNN model was better than the ANN and SVM models as it outperformed them on all points.The CNN was deployed in a user-friendly security architecture based on HCI concepts to make it easy to use without compromising security.User answers indicated excellent satisfaction (4.7), responsiveness (4.8), and trust (4.9), along with a false alarm rate of 2.1%, showing the framework's security usability and dependability.The study reveals that the CNN is able to detect threats with good accuracy.Also, it shows how user design helps generate trust and compliance.The study reveals considerable promise in the usage of CNNs in cyber-security.The researchers note that despite employing only a single dataset and the com-laxity of CNN models, the findings illustrate the significance to the future.They say it enables the opportunity to construct continuous HCI-ML convergence in cybersecurity.This can lead to the establishment of durable, trustworthy, and userfriendly digital places.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.