An Investigation of Various Techniques to Improve Cyber Security
Shoaib Mohammad, Ramendra Pratap Singh, Rajiv Kumar, Kshitij Kumar Rai, Arti Sharma, Saloni Rathore · 2025
Cyber security threats, characterized by a series of assault stages, persistently aim to accomplish a pre-established goal. Due to the intricate nature of these attacks, the intruder is capable of bypassing the target's security defenses and gaining access to a majority of its systems. When accessing data kept in the cloud, there is a genuine risk of experiencing data breaches, compromised credentials, Denial of Service (DoS) assaults, hacked interfaces and Application Programming Interfaces (APIs), permanent data loss, and other significant cybersecurity concerns. Due to the constant innovation of cybercriminals, who continuously develop more advanced methods to avoid detection, it is challenging to both identify and prevent these malicious activities. As digital technology advances, gigabytes and terabytes of data are now generated every second. Businesses in a variety of industries are finding that using the internet to manage their resources and transactions is useful. Given the value of data and the need to safeguard its security and privacy, securing big data remains a major challenge for all solutions. Due to the exponential expansion of network data, intrusion detection is becoming increasingly important, and manual analysis would be either impossible or take the same amount of time as analyzing it. As a result, there is an urgent need for an automated system capable of extracting relevant information from enormous amounts of hitherto untapped data when it comes to network intrusion detection. Data mining can perform a variety of tasks, such as clustering, prediction, classification, and the extraction of association rules between data pieces. This paper discusses machine learning techniques for designing intrusion detection systems for big data networks. In this approach, the NSL KDD data set is used as input. First, the CFS-correlation feature selection approach is used to pick only relevant features from the NSL KDD data set. The NSL KDD data collection contains 41 features. The number of characteristics was reduced to 16 after applying the CFS algorithm. The 16 attributes are then used by machine learning techniques to classify and predict malware data in the NSL KDD data set.