A Modified Regression Model for Analysing the Performance of Metamaterial Antenna Using Machine Learning and Deep Learning

Metamaterial (MM) is an artificial constituent, which as a distinct properties i.e., negative permittivity and refractive-index, which doesn’t subsists naturally in the environment. MM is widely used in the antenna application owing to their abundant advantages. Split Ring-Resonator configuration in MM antenna can enhance the performance of antenna. Therefore, the present study aims to develop an improved regression model, which evaluates the performance of MM antennas effectively. In that context, initially, the study performs, pre-processing by removing the unnecessary data and missing values. Then, the feature extraction is employed with Bi-LSTM, which extracts the efficient features. Followed by this, the training and testing split is performed as 80% training data and 20% of testing data. The modified regression model is constructed with empirical loss function and XG-Boost algorithms. By implementing the proposed model, the prediction phase is enhanced efficiently. The experimental evaluation of the proposed model is accomplished with Metamaterial antenna dataset (MM antenna dataset), which is openly available and the analysis is done in terms of error rate and accuracy. The proposed system attains 99.23% accuracy rate.