Convolutional Neural Networks for Intrusion Detection in IoT Networks with Limited Resources

In this research, we explore the adequacy of Convolutional Neural Systems (CNNs) for interruption location in resource-constrained IoT systems. Leveraging two freely accessible datasets, we compare the execution of CNNs with conventional machine learning calculations such as Support Vector Machines (SVM) and Random Forests (RF). Our tests uncover that CNNs accomplish competitive location exactness, beating SVM and RF models over both datasets. Particularly, on Dataset A, CNNs accomplish an accuracy of 0.92, a review of 0.88, and an F1-score of 0.90, whereas, on Dataset B, they accomplish an accuracy of 0.86, a review of 0.84, and an F1-score of 0.85. Also, CNNs illustrate prevalent computational proficiency, with shorter preparing times and lower memory impressions compared to conventional calculations. This inquiry contributes to progressing the state-of-the-art in IoT interruption discovery by highlighting the viability of profound learning methods in tending to cybersecurity challenges in IoT situations. Moving forward, the selection of CNNs can upgrade the versatility and security of IoT systems, clearing the way for more strong and effective interruption location frameworks.