Secure Sensitive Information on IoT using Machine Learning

• Published in: 2023 International Conference on Sustainable Communication Networks and Application (ICSCNA) pp. 360 - 366
• Main Authors: Nagendiran, S., Renugadevi, R., Kumar, R. P. Anto, Sasirekha, K., Harini, R.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 15-11-2023
• Subjects: Artificial Intelligence; Medical services; Multiclass Classification; Natural language processing; Production; Random Forest Classifier; Reinforcement learning; Risk management; Support vector machine classification; Telecommunication traffic
• DOI: 10.1109/ICSCNA58489.2023.10370157

In a number of industries, including banking, cybersecurity, healthcare, and others, risk assessment is an essential procedure. By automating data analysis, seeing trends, and offering insights that might help organisations make better decisions, artificial intelligence (AI) tools can greatly improve risk assessment. Here are some typical AI methods and strategies for risk assessment. Machine learning is the first step, followed by Natural Language Processing (NLP), Reinforcement Learning, Predictive Analytics, and so on. While AI may significantly improve risk assessment, it ought to be used in combination with human judgement and domain knowledge. Additionally, when using AI for risk assessment, it's imperative to address issues with the accuracy of data, bias, and ethics. To guarantee the accuracy and continued relevance of risk evaluations, regular verification of models and updating is also necessary. The importance of the attributes in the UNSW-NB15 collections is therefore thoroughly analysed in this work. This research analyses a UNSW-NB15 collection of data generation to address the problems associated with the lack of any network reference data sets. This data collection combines network traffic assault actions that are now synthesised with real-world modern norms. The UNSWNB15 information set's characteristics are produced using both established and cutting-edge techniques. Many Machine Learning algorithms are used to analyse the intrusion by measuring the accuracy and other features. Here the data set is classified as Binary classification and Multi Class Classification where Machine learning are applied to find out the intrusion. Random Forest Classifier (RFC) performance good for binary classification and Linear Support Vector Machine(LSVm)performance is good for Multiclass Classification.