부산대학교 도서관

Fuzzy logic, machine learning, and artificial intelligence (AI) together provide a ground-breaking approach to increased security. This change in perspective creates new opportunities that may aid in identifying vulnerabilities and strengthening protocols, networks, and software. Fuzzer activities and other anomalies in a network may be easily identified by machine learning because to its ability to analyze historical data and observe events in real time. Future-facing features like as sophisticated threat models, artificial intelligence (AI)-powered automated responses, and the integration of threat intelligence will transform vulnerability discovery. In order to strengthen defensive efforts, collaboration amongst other areas, adherence to safety regulations, and user education will be necessary. The promising combination of machine learning and fuzzer identification might make the internet a safer place since cyber dangers are always evolving. In this research paper, a systematic approach is employed for early detection of fuzzer attack using UNSW-NB15 dataset. The research work’s proposed method archives noteworthy performance of finding changes from how the network usually works and quickly letting admins know about any possible security vulnerabilities or threats. Because of this early notice, companies can take steps to protect their digital assets and make it less likely that they will be misused. The experiments are performed to evaluate the capability of logistic regression, gaussian naïve bayes and LSTM machine learning techniques on dataset for fuzzer attack and benign traffic patterns and comparative anlyais are presented in the research paper.