Privacy-Preserving Deep Learning for Detecting Zero-Day Smishing Attacks in Digital Networks

Cell phones are so widely used and always connected to the Internet, they are susceptible to phishing and smishing crimes. The practice of sending users mistaken short messages is known as phishing. Integrating the methods of phishing with SMS, smishing involves attackers sending the victim an SMS with malicious content. SMS communications are rapidly growing in the twenty-first century. SMS is among the most significant communication techniques accessible today. Every day, billions of SMS are exchanged worldwide, increasing people's vulnerability to threats. From our electronic devices, spam SMS have been used for extortion, blackmail, and phishing. While recognizing email spam is a standard practice, detecting SMS spam us-ing the Bengali dataset is a novel research challenge. This study uses both Bi-LSTMS with Attention Mechanism and conventional machine learning tech-niques to find Bengali spam SMS. The best algorithm is then figured out by comparing the performances of all the algorithms. At 99%, Bi-LSTMS with Attention Mechanism gets the best testing accuracy rate.