Innovative Method for Earthquake Prediction System using Hybrid Convolutional Neural Network and SVM

Abstract

The ability to estimate casualties from earthquakes is crucial for effective disaster response. Conventional forecasting techniques have stringent sample data requirements and several parameters that must be manually specified, which can lead to subpar outcomes with prediction accuracy as low and a slow rate of learning. In the suggested hybrid model, CNN is employed as an automatic feature extractor, while SVM is used as a binary classifier. Traditional CNN’s completely linked layers are swapped out for a support vector machine in this model to improve prediction accuracy. This proposed approach employs CNN for automatic feature extraction, and an SVM classifier for automatic classification. The experimental findings showed that compared to the CNN model (89%), our hybrid model was significantly more accurate at 98.5%