Hyper-parameter Optimization of LSTM Network Using Genetic Algorithm and Q-Learning Algorithm for Classification of COVID-19 Dataset

Currently, a lot of worry and anxiety among people both locally and internationally about the coronavirus potentially becoming a pandemic and spreading globally. However, in recent years, more and more people have been relying on social networks as their primary source of news and information. As a result, many political figures are greatly concerned about the widespread false and deceptive information on social media. We are not only facing the challenge of combating COVID-19 but also an "infodemic." To tackle this issue, A dataset consisting of 7,000 social media posts in Persian related to COVID-19 has been collected and made available. This data consists of both true and false news. Several languages, such as Arabic, English, Chinese, and Hindi, have also recognized the problem of fake news related to COVID-19. In this study, a deep neural network approach is used to simplify feature extraction, develop a strong ability to learn, and automatically discover features, which is more effective than traditional machine learning approaches. Additionally, a novel approach to improving outcomes using a deep neural network is employed. The hyperparameters of the deep learning algorithm are set and optimized using genetic algorithms and Q-learning, resulting in better outcomes than previous research and achieving an accuracy rate of 0.92 percent.