Hybrid ANN and Ant Colony Algorithm for IoT Data Classification

In this study, a hybrid approach based on the Ant Colony Optimization (ACO) algorithm and a Multilayer Perceptron (MLP) neural network is proposed for classifying Internet of Things (IoT) data. Managing and analyzing the large volume of data in IoT networks with multiple sensors presents a significant challenge. This research employs the ACO algorithm to identify and select the most relevant sensors and features from the collected data. By performing a parallel search through the feature space, ACO selects the best feature combinations from the vast amount of data, resulting in more accurate classification. These optimized features are then used to train the MLP neural network. The combination of ACO and MLP improves both classification accuracy and the learning speed of the model. ACO enhances model performance by efficiently optimizing the features and avoiding local optima. Meanwhile, the MLP's multilayer structure strengthens its ability to learn complex patterns and recognize non-linear relationships in IoT data. Experimental results show that ACO alone achieves an accuracy of ۰.۸ and an F۱-Score of ۰.۸۳۹, while the combination with MLP increases the accuracy to ۰.۸۱ and the F۱-Score to ۰.۸۴۹. This improvement indicates that integrating ACO with MLP enhances classification precision and strengthens the model's ability to learn complex patterns. The hybrid approach also reduces processing time and adapts better to dynamic, large-scale IoT data, demonstrating that combining swarm intelligence methods like ACO with neural networks provides an effective solution for IoT data processing and classification.

Keywords:

ant colony optimization algorithm , artificial neural network , data classification , internet of things