همایش بین المللی هوش مصنوعی

صفحه اصلی / اولین همایش بین المللی هوش مصنوعی

Enhancing IoT Data Prediction Accuracy Using Deep Learning and Metaheuristic Algorithms

نویسندگان :

Safoura Ashoori¹ Khadigh Nemati² Mohamad hadi Amini³

1- دانشکده فنی میرزا کوچک صومعه سرا 2- دانشکده فنی حرفه ای میرزا کوچک صومعه سرا 3- دانشکده فنی حرفه ای میرزا کوچک صومعه سرا

کلمات کلیدی :

ffn

چکیده :

Given the increasing volume of data generated by the Internet of Things and the challenges associated with processing and storing this data in cloud environments, it is essential to employ deep learning methods and metaheuristic algorithms to improve the accuracy of stream data prediction. In this study, four different approaches were evaluated for classifying continuous IoT data: Particle Swarm Optimization, Support Vector Machine, the PSO-SVM combination, and a feedforward neural network integrated with PSO. Considering the characteristics of stream data and the need to avoid local optima, the PSO algorithm was utilized to optimize the weights and parameters of the feedforward neural network. Additionally, PSO was combined with SVM to optimize its parameters, achieving an accuracy of 0.71. The combination of FFN with PSO improved the prediction accuracy to 0.73, demonstrating the superior performance of this method compared to others. These results highlight the high potential of combining deep learning and metaheuristic methods in enhancing the classification accuracy of IoT data