Abstract:
Machine learning is super important in cybersecurity. The main goal of using machine
learning in cybersecurity is to improve how we detect malware. This makes the process
doable, scalable, and way more effective than old-school methods that need people to
step in. Dealing with machine learning challenges in cybersecurity needs some serious
handling. Different methods like deep learning, support vector machines, and Bayesian
classification have shown they work well in stopping cyber-attacks. It's crucial to find
hidden insights from network data and use a data-driven machine learning model to
stop these attacks before they happen. No research has been done to understand how
vulnerable ML techniques are against security threats and how they can be defended
against. We need researchers, scientists, and engineers to start focusing on
cybersecurity with ML. This survey looks at the machine learning techniques used on
cybersecurity data to secure systems. However, ML can be attacked during training and
testing, causing problems with security. We also talk about current cybersecurity threats
and how machine-learning techniques help fight them off. We point out the flaws in
these fancy models and how attack tactics have changed over the years. We aim to see
if these machine-learning techniques work against the growing problem of malware
causing trouble online.
