Evasive SMS spam detection using machine learning models
T. N. Ranganadham, Bathula Mounika, D. K. Meghana, Shaik Mastan Vali, K. Janardhan · 2025
The sophistication of SMS spam, which is fueled by spammers’ use of evasive strategies, is becoming a bigger issue for detection systems. This study examines and assesses machine learning techniques for evasive SMS spam detection in order to improve spam detection efficiency. A recently curated dataset of over 72581 SMS messages—including both spam and legitimate (ham) communications—is used to evaluate the efficiency of various models. They include conventional machine learning models such as Support Vector Machines, Random Forests, Decision Trees, and Multinomial Naive Bayes. To find out how well these models can resist spammers’ evasion strategies, we assess their resilience to adversarial and obfuscation attacks.