IDENTIFICATION OF ANOMALOUS CYBER ACTIVITY PATTERNS IN IOT DATA STREAMS

Abstract

Artificial intelligence (AI) is progressively urgent within the advancement of vigorous cybersecurity arrangements. The need of security investigation is crucial as our digital landscape develops. System security is provided by Arrange Interruption Location Systems (NIDS), which swiftly locate and stop arranged breaches. With the appearance of cutting-edge machine learning strategies, especially imaginative neural organize plans, the adequacy of interruption discovery has altogether moved forward. This consider assesses the execution of contemporary machine learning techniques employing a special cybersecurity benchmark dataset custom-fitted for IoT applications, known as IoT-23. Particularly, we use the capabilities of Profound Autoencoder (DAE) for effective dimensionality decrease. Furthermore, a suite of machine learning strategies, enveloping Profound Neural Systems and long-term and short-term memory (LSTM). systems is utilized to distinguish between ordinary and malevolent arranged designs. The viability of our approach is thoroughly tried and approved utilizing the IoT-23 dataset. At last, the results of this examination are fastidiously scrutinized and translated based on different assessment measurements.