Your browser doesn't support javascript.
Analyzing Deep Learning Optimizers for COVID-19 Fake News Detection
Studies in Computational Intelligence ; 1001:401-420, 2022.
Article in English | Scopus | ID: covidwho-1592202
ABSTRACT
In this time of COVID-19 crisis, the threat posed by the propagation of misinformation leading to mistrust needs to be kept in check. Misinformation related to the vaccines, remedies, false symptoms, etc. are spiraling out of control. We might not be able to directly put a stop to the flow or spread of fake news to a large extent at the moment, but it may be able to identify it as such with the help of Natural Language Processing (NLP) tools and Deep Learning (DL) algorithms. Steps involved in achieving this goal can be narrowed down into collection and analysis of data from various sources, sorting out the articles as covid-relevant and categorizing them as real or fake using DL models. However, DL models use different optimizers in the learning process, which plays an important role in identifying the fake news. This chapter also compares the efficiency of different optimizers in the context of COVID-19 fake news detection using DL models. The newly developed Continuous Coin Betting (CoCoB) Optimizer for DL studied extensively for fake news detection and performed compared with four other widely used optimizers. The comparative analysis shows the CoCoB as well as popularly used Adam optimizers are quite effective in finding optimal classification results for detection of fake news related to COVID-19. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
Keywords

Full text: Available Collection: Databases of international organizations Database: Scopus Language: English Journal: Studies in Computational Intelligence Year: 2022 Document Type: Article

Similar

MEDLINE

...
LILACS

LIS


Full text: Available Collection: Databases of international organizations Database: Scopus Language: English Journal: Studies in Computational Intelligence Year: 2022 Document Type: Article