Analysis of Malware Attacks in Banking, Financial Services, and Insurance (BFSI) Using Deep Denclue Regression-EM

Nowadays, one of the primary concerns of banks is cyber security. Clients are becoming increasingly digital, exposing them to greater hacking dangers. Wide files including data on company procedures, customer records, and all secret evidence can be deleted if nothing is done to safeguard all of this. A cyber-security breach may have major consequences for both personal and commercial consumers, in addition to destroying credibility. The need for IoT devices is fast increasing in the Banking, Financial Services, and Insurance (BFSI) sector, and their role in everyday life is also referred to as smart gadgets. However, cyber thieves perceive money opportunities, which causes them to intensify and differentiate their assaults. The risk for consumers who like Internet of Things (IoT) gadgets is that threats might be created unexpectedly and that seemingly harmless methods can become authoritative tools for criminal activities. These might include malicious crypto money withdrawal, DDoS assaults, or botnet activities requiring computer disclosure. When malware infects the victim's IoT system, it takes control of the victim's PC and performs harmful operations. Linear Deep DENCLUE (DENsity CLUstEring) (Grouping Built on Corpus Spreading Meaning) Regression(LDR) clustering procedure with Expectation Maximization(EM) to dynamically cluster the IoT executable Application Programming Interface (API) calls, kernel service calls, and network-related kernel API calls. The pool network used LDR to link unknown executable API requests that were executing Victim device harmful activities on BFSI (Banking, Financial Services, and Insurance) servers communicating with IoT devices. Once the clustered API call network connected with the IoT kernel was constructed, the Expectation-Maximization technique was used to find the closest proximity to harmful behaviour. The suggested work was evaluated with 1052 malware samples from various malware families, yielding an accurate optimistic of 97.47 percent and an inaccurate optimistic of 0.09 percent.