AN INTELLIGENT AND SECURE SPAM CLASSIFICATIONFRAMEWORK BASED ON SUPERVISED LEARNING

Abstract

Numerous sensors and actuators are connected through wired or wireless paths to enable data transmission. This field has seen rapid growth over the past decade, and by 2020 it was expected to be connected to more than 25 billion devices. In the next five years, these devices are expected to transmit significantly larger volumes of data than they do currently. Devices generate large amounts of data in multiple formats, with data quality influenced by factors such as the source of generation, processing time, and data volume. Machine learning techniques play a vital role in ensuring security and access control in biotechnology, while also enhancing safety and efficiency by identifying abnormal behavior. At the same time, malicious users often employ learning techniques to exploit system vulnerabilities. After analyzing these challenges, we propose the use of machine learning methods to identify and reduce spam, thereby improving device security. The most effective way to detect waste is through an autonomous learning system. This approach evaluates the performance of four machine learning models using different feature sets and metrics as inputs. All models utilize updated input attributes to calculate a spam score. This score, influenced by multiple factors, reflects the reliability of the device. The results show that the proposed method performs effectively when compared with other well-established systems.