Resource-Efficient Hybrid Machine Learning Model for IoT SMS Spam Detection

The rapid growth of text-based communication in IoT devices has led to the increase of SMS spam messages. This project presents a resource-efficient hybrid machine learning model aimed at classifying SMS messages into two categories: "ham" and "spam." The approach combines Bi-GRU with CatBoost, Bi-GRU with AdaBoost, and BERT with XGBoost to create a model that efficiently detects spam while ensuring high classification accuracy. The dataset used for training contains SMS messages labeled as "ham" or "spam," each containing a text message. The proposed model is designed to reduce computational resources while achieving high performance in text classification tasks. The backend is developed using the Flask framework, while the frontend is built using HTML, CSS, and JavaScript. The combination of different machine learning techniques allows the model to handle text classification effectively with minimal resource consumption, making it suitable for IoT environments where computational power may be limited. The project focuses on optimizing both model efficiency and classification accuracy for practical applications in automated SMS filtering.

Keywords: Hybrid model, Bi-GRU, CatBoost, AdaBoost, BERT, XGBoost, SMS classification, IoT, Spam detection, Resource efficiency.