Tuning the hyperparameters for supervised machine learning classification, to optimize detection of IoT Botnet

We all agreed that we are living in the Internet of Things (IoT) generation, pretty much any physical object that exists can transform into an IoT device if it can be connected or controlled to exchange information with a given network. That means in cybersecurity terms a lot of opportunities to create IoT botnets to break down the networks of these devices. Knowing that the majority of these devices are usually quite cheap, and of course, mass market-oriented, with no attention paid to access control, data protection, or any further security management. Consequently, they are immensely the favorite target for hackers to exploit their vulnerabilities to construct very many types of IoT botnets with different characteristics and goals, beginning by spying on a network to get control of all the other devices. For that, this paper comes to propose a study about how we can exploit the powerful benefits of supervised machine learning classification algorithms and models by tuning their hyperparameters, in order to regularize and minimize the measures of the error to get in last the best detecting of any type of these kinds of the botnet on a given network. This paper will combine two powerful innovations in computer science IoT and machine learning to give a hand of help on one of the most complicated fields in this science which is IoT network security by finding the answer of which is the favorable sets of hyperparameters that can be tuned in order to optimize the performance of detecting IoT botnet using some supervised machine learning classification models.