BERT-Based Hybrid RNN Model for Multi-class Text Classification to Study the Effect of Pre-trained Word Embeddings

Due to the Covid-19 pandemic which started in the year 2020, many nations had imposed lockdown to curb the spread of this virus. People have been sharing their experiences and perspectives on social media on the lockdown situation. This has given rise to increased number of tweets or posts on social media. Multi-class text classification, a method of classifying a text into one of the pre-defined categories, is one of the effective ways to analyze such data that is implemented in this paper. A Covid-19 dataset is used in this work consisting of fifteen pre-defined categories. This paper presents a multi-layered hybrid model, LSTM followed by GRU, to integrate the benefits of both the techniques. The advantages of word embeddings techniques like GloVe and BERT have been implemented and found that, for three epochs, the transfer learning based pre-trained BERT-hybrid model performs one percent better than GloVe-hybrid model but the state-of-the-art, fine-tuned BERT-base model outperforms the BERT-hybrid model by three percent, in terms of validation loss. It is expected that, over a larger number of epochs, the hybrid model might outperform the fine-tuned model. © 2022,International Journal of Advanced Computer Science and Applications. All Rights Reserved.

Ensemble Model to Forecast the End of the Covid-19 Pandemic

The coronavirus disease 2019 (Covid-19) epidemic has caused a worldwide health catastrophe that has had a profound influence on how we see our planet and our daily lives. In this pandemic circumstance, machine learning (ML) based prediction models demonstrate their value in predicting perioperative outcomes to enhance decision-making on future course of action. Ensemble learning is used in the majority of ML based forecasting approaches. The ML models anticipate the number of patients who will be affected by Covid-19, and use this information to forecast the end of the pandemic is to be leveraged. Three types of predictions are made: the number of newly infected cases, the number of deaths, and the number of recoveries in the next ‘x’ number of days. By combining one of the forecasting models with classifiers, we can predict the end of the pandemic. The proposed idea combines the SIRF model from epidemiology and a forecasting machine learning model named Prophet and a Naïve Bayes Classifier to predict the end of the pandemic. Using the theoretical equations of the SIRF model, we developed a formula for infectious growth rate. The classifier uses this infectious growth rate to check if the infection is fading. With confirmed, recovered and fatalities data, the infectious growth rate is calculated. Naïve Bayes classifier is used to check if the pandemic is about to end or not. If not then forecast the data for ‘x’ number of days and do the calculations again. The process continues until we get a time frame where the pandemic may reach its end. The results are discussed for 2 countries India and Israel. The forecasts done for Israel were very accurate to the actual data, whilst for India it was less comparatively as India was hit by 2 waves of Covid-19 pandemic. By leveraging the forecasting and classification capabilities of machine learning models like FBProphet, Naïve Bayes Classifier, and the mathematical equations of the SIRF model from epidemiology, the life span of the pandemic is determined. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

The Face Mask Detection Technology for Image Analysis in the Covid-19 Surveillance System

Face mask recognition has been growing rapidly after corona insistent last years for its multiple uses in the areas of Law Enforcement Security purposes and other commercial uses Face appears spreading others to corona a novel approach to perform face new line detection and face mask recognition is proposed. The proposed system to classify face mask detection using COVID-19 precaution both in images and videos using convolution neural network. Extensive experimentation on the datasets and the performance evaluation of the proposed methods are exhibited. Further, we made a successful attempt to preserve inter and intra class variations of face mask detection using symbolic approach. We studied the different classifiers like Support Vector Machine and a Symbolic Classifier. The project is developed as a prototype to monitor temperature measurement and to detect mask for the people. The first method is performed using temperature sensor used to detect the present temperature of the body and automatically spray the sanitizer. In the second method, the work is designed to provide a safety system for the people in order to avoid COVID-19. We proposed continuous monitoring of the people conditions and store the people’s data in the server using the Deep learning concept. In order to investigate the performance the proposed method an extensive experimentation is conducted on 50 various Image dataset. We conducted experimentation under varying of training and testing percentage for 10 random trails. From the results we could observe that, the results obtained for symbolic approach is better than the conventional approach.