Reinforcement Learning-based Counter-Misinformation Response Generation: A Case Study of COVID-19 Vaccine Misinformation

The spread of online misinformation threatens public health, democracy, and the broader society. While professional fact-checkers form the first line of defense by fact-checking popular false claims, they do not engage directly in conversations with misinformation spreaders. On the other hand, non-expert ordinary users act as eyes-on-the-ground who proactively counter misinformation - recent research has shown that 96% counter-misinformation responses are made by ordinary users. However, research also found that 2/3 times, these responses are rude and lack evidence. This work seeks to create a counter-misinformation response generation model to empower users to effectively correct misinformation. This objective is challenging due to the absence of datasets containing ground-truth of ideal counter-misinformation responses, and the lack of models that can generate responses backed by communication theories. In this work, we create two novel datasets of misinformation and counter-misinformation response pairs from in-the-wild social media and crowdsourcing from college-educated students. We annotate the collected data to distinguish poor from ideal responses that are factual, polite, and refute misinformation. We propose MisinfoCorrect, a reinforcement learning-based framework that learns to generate counter-misinformation responses for an input misinformation post. The model rewards the generator to increase the politeness, factuality, and refutation attitude while retaining text fluency and relevancy. Quantitative and qualitative evaluation shows that our model outperforms several baselines by generating high-quality counter-responses. This work illustrates the promise of generative text models for social good - here, to help create a safe and reliable information ecosystem. The code and data is accessible on https://github.com/claws-lab/MisinfoCorrect. © 2023 Owner/Author.

Night-time Human Mobility during Pandemic in the Philippines as Observed by VIIRS Satellite

Recent research has analyzed and studied the growing literature on human mobility during quarantine periods using various methodology and techniques. There are several ways to use light pollution to assess mobility. The data from the VIIRS satellite can be used to quantify light pollution and human mobility in the Philippines during quarantine. The data utilized in this study came from NASA's EOSDIS Worldview website. The number of cases and pixels count increases from early April 2020 to late August 2020. However, the cases increased from February to April 2021. This could be attributed to the active human mobility seen between December 2020 and January 2021. Human interactions have been intense since August 2020, causing an increase in COVID cases that peaked between March and April 2021, before dropping in May 2021. Following the conclusion of this study, light pollution VIIRS satellite pictures can be used to identify possible COVID- 19 cases. There are many more factors and variables to consider when writing a comprehensive paper. With the relaxed quarantine time has been achieved beyond June 2021, additional dates may be explored since there may be a direct relationship between light pollution and COVID-19 instances. © 2022 IEEE.

I am what I am-Convergence Behaviors on Online Discussion about the Safety of COVID-19 Vaccines

Using data from an online discussion on the risk of getting blood clot from Johnson & Johnson vaccine moderated by the New York Times Facebook page, we investigated the presence of eleven convergence behaviors, and the interaction between them. While recent research focuses on misinformation or fake news as the object of analysis, we argue in this exploratory research that it is equally important to analyze who and, whenever possible, why people engage in information exchange given a particular crisis, hence their convergence behaviors. Mapping the types of postings to their authors would be an additional step to design, develop, implement, and possibly, regulate online discussions for a more effective and just civic engagement. As we witness a mass manipulation of public opinion, our findings suggest that the number of netizens that seek to correct misinformation is growing. If the society goal is to swiftly rebut as many conspiracy theories as possible, we advocate for a dual social media control strategy: restrain as much as possible the misinformation spreaders/manipulators and encourage correctors to help propagate countervailing facts. © 2022 IEEE Computer Society. All rights reserved.

Contextualizing Emerging Trends in Financial News Articles

Identifying and exploring emerging trends in news is becoming more essential than ever with many changes occurring around the world due to the global health crises. However, most of the recent research has focused mainly on detecting trends in social media, thus, benefiting from social features (e.g. likes and retweets on Twitter) which helped the task as they can be used to measure the engagement and diffusion rate of content. Yet, formal text data, unlike short social media posts, comes with a longer, less restricted writing format, and thus, more challenging. In this paper, we focus our study on emerging trends detection in financial news articles about Microsoft, collected before and during the start of the COVID-19 pandemic (July 2019 to July 2020). We make the dataset accessible and we also propose a strong baseline (Contextual Leap2Trend) for exploring the dynamics of similarities between pairs of keywords based on topic modeling and term frequency. Finally, we evaluate against a gold standard (Google Trends) and present noteworthy real-world scenarios regarding the influence of the pandemic on Microsoft. ©2022 Association for Computational Linguistics.

Identification and Localization COVID-19 Abnormalities on Chest Radiographs

Solutions to screen and diagnose positive patients for the SARS-CoV-2 promptly and efficiently are critical in the context of the COVID-19 pandemic's complex evolution. Recent researches have demonstrated the efficiency of deep learning and particularly convolutional neural networks (CNNs) in classifying and detecting lung disease-related lesions from radiographs. This paper presents a solution using ensemble learning techniques on advanced CNNs to classify as well as localize COVID-19-related abnormalities in radiographs. Two classifiers including EfficientNetV2 and NFNet are combined with three detectors, DETR, Yolov7 and EfficientDet. Along with gathering and training the model on a large number of datasets, image augmentation and cross validation are also addressed. Since then, this study has shown promising results and has received excellent marks in the Society for Imaging Informatics in Medicine's competition. The analysis in model selection for the trade-off between speed and accuracy is also given. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Opening up Minds with Argumentative Dialogues

Recent research on argumentative dialogues has focused on persuading people to take some action, changing their stance on the topic of discussion, or winning debates. In this work, we focus on argumentative dialogues that aim to open up (rather than change) people's minds to help them become more understanding to views that are unfamiliar or in opposition to their own convictions. To this end, we present a dataset of 183 argumentative dialogues about 3 controversial topics: veganism, Brexit and COVID-19 vaccination. The dialogues were collected using the Wizard of Oz approach, where wizards leverage a knowledge-base of arguments to converse with participants. Open-mindedness is measured before and after engaging in the dialogue using a questionnaire from the psychology literature, and success of the dialogue is measured as the change in the participant's stance towards those who hold opinions different to theirs. We evaluate two dialogue models: a Wikipedia-based and an argument-based model. We show that while both models perform closely in terms of opening up minds, the argument-based model is significantly better on other dialogue properties such as engagement and clarity. © 2022 Association for Computational Linguistics.

Prompt Learning for Multi-modal COVID-19 Diagnosis

The outbreak of COVID-19 pandemic has spread rapidly and severely affected all aspects of human lives. Recent researches has shown artificial intelligence and deep learning based approaches have achieved successful results in detecting diseases. How to accurately and quickly detect COVID-19 has always been the core topic of research. In this paper, we propose a novel approach based on prompt learning for COVID-19 diagnosis. Different from the traditional 'pre-training, fine-tuning' paradigm, we propose the prompt-based method that redefine the COVID-19 diagnosis as a masked predict task. Specifically, we adopt an attention mechanism to learn the multi-modal representation of medical image and text, and manually construct a cloze prompt template and a label word set. Selecting the label word corresponding to the maximum probability by pre-training language model. Finally, mapping the prediction results to the disease categories. Experimental results show that our proposed method obtains obvious improvement of 1.2% in terms of Mi-F1 score compared with the state-of-the-art methods. © 2022 IEEE.

Deep Learning-Based COVID-19 Screening Using Photographs of Chest X-Rays Displayed in Computer Monitors

Several recent research papers have shown the usefulness of Deep Learning (DL) techniques for COVID-19 screening in Chest X-Rays (CXRs). To make this technology accessible and easy to use, a natural path is to leverage the widespread use of smartphones. In these cases, the DL models will inevitably be presented with photographs taken with such devices from a computer monitor. Thus, in this work, a dataset of CXR digital photographs taken from computer monitors with smartphones is built and DL models are evaluated on it. The results show that the current models are not able to correctly classify this kind of input. As an alternative, we build a model that discards pictures of monitors such that the COVID-19 screening module does not have to cope with them. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

An Empirical Model of Classifying Lung Affected Diseases to Detect COVID-19 Using Chest X-ray Employing Convolutional Neural Architecture

The earlier detection and accurate diagnosis of COVID seem to be a global problem. It is difficult to make a large number of testing equipment, but then again, their reliability is relatively poor. Recent research indicates the usefulness of chest x-ray pictures in identifying COVID. This study presents a deep learning algorithm developed from the ground up to categorize as well as confirm the existence of COVID in a set of X-ray imaging data. We designed a CNN architecture from the ground up to retrieve elements from provided X-ray data to categorize them and identify the individual contaminated with COVID. Our strategy may aid in mitigating the consistency issues while working with medical data. In contrast to some other classifying activities with a large enough image database, obtaining large X-ray datasets for this classification job is challenging. So, we applied multiple data enhancement techniques to maximize the accurateness, achieving a significant accuracy of 97.75 percent. © 2022 IEEE.

Analysis of Spreading Fake Corona News Using Python

Fake news has flourished for quite some time.Fake news is spreading at a rapid rate due to the rapid growth of smartphone users and the ease of access to the internet. Is there a reason why so many people are so willing to believe bogus news? How come we don't double-check our information before passing it on? Other questions remain unresolved, such as: The panic caused by the spread of false information during the crisis. It took a long time for the Covid-19 pandemic to break out. In this research, we used the dataset Fake.csv to examine several scenarios of fake news spreading in different continents, countries, and age groups. There has been a lot of recent research on the topic. © 2022 IEEE.

Capsule Networks with Chest X-Ray Enhancement for Detection of COVID-19

Corona virus was declared a global pandemic that has affected people worldwide. It is critical to diagnose corona virus-infected individuals to restrict the virus's transmission. Recent research indicates that radiological methods provide valuable information in identifying infection using deep learning algorithms. Deep learning has contributed to large-scale medical data research, providing new ways and chances for diagnostic tools. This research attempted to investigate how the Capsule Networks leverage chest X-ray scans to identify the infected person. We suggest Capsule Networks identify the illness using chest X-ray data. The proposed approach is rapid and robust, classifying scans into COVID-19, No Findings, or any other issue in the lungs. The study can be used as a preliminary diagnosis by medical practitioners, and the study focuses on the COVID-19 class, a minority class in all public data sets accessible, and ensures that no COVID-19 infected individual is identified as Normal. Even with a small dataset, the model provides 96.37% accuracy for COVID-19 and for the non-COVID-19, and on multi-class classification, it provides an accuracy of 95.12%. © 2022 IEEE.

Say No2Ads: Automatic Advertisement and Music Filtering from Broadcast News Content

The incredible growth in available news content has been met with steeply increasing demand for news amongst the general population. The 24/7 news cycle gives people an awareness of events, activities and decisions that may have an impact on them (e.g. the latest updates on the COVID-19 outbreak). Despite the flourish of social networks, recent research suggests radio and especially TV are still the main sources of news for many people. However, unlike in social media, the content aired on radio and TV requires people to listen to every single advertisement and music (for radio) before consuming the next item. For this reason, media monitoring companies have to dedicate considerable amount of resources on processing or manually filtering the advertising content (which is blended with the actual news). Often their clients still receive ads. To mitigate this problem, in this paper, we propose No2Ads, an autoregressive deep convolutional neural network (CNN) model that is trained on over 500 h of human annotated training samples to remove ads and music from broadcast content. No2Ads reached very high performance results in our tests, achieving 97% and 95% in precision and recall on detecting ads/music for radio channels;95% precision and 98% recall for TV channels. Between March to September 2021, across 261 radio and TV channels in Australia and New Zealand, No2Ads has detected and filtered out 22,161 h of all captured broadcast content as either advertisements or music. © 2022, Springer Nature Switzerland AG.

Forecasting of COVID-19 Trends in Bangladesh Using Machine Learning Approaches

Bangladesh, a low-to-middle income economy country with one of the world’s densest populations, is ranked 26th worldwide having a positive case rate of 25%-30% of COVID-19 confirmed instances as of July 28, 2021. The recent researches related to COVID-19 focus on addressing mental health problems caused by it, and fewer works have been performed to forecast its trends using machine learning (ML), especially in Bangladesh. Therefore, this research attempts to predict the infected, death, and recovery cases for COVID-19 in Bangladesh using four ML techniques FB Prophet, ARIMA, SARIMAX, LSTM and com pare their forecasting performance to find out the best prediction model. The experimental results showed that for ‘Detected’ and ‘Death’ case, LSTM and SARIMAX performed better than other models with (RMSE = 1836.79, MAE = 1056.36) and (RMSE = 24.70, MAE = 15.54), respectively. In the ‘Recovery’ case, the best result was found in the ARIMA model with RMSE = 558.87, MAE = 299.64. According to the analysis, this research work can help predict the trends of COVID-19 cases in the future and help policymakers taking necessary precautions to control the detection and death rate in the country. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

A Survey for Using AI Techniques for Predicting COVID-19

Artificial intelligence has finally brought in a qualitative leap in health care, allowing for the exploration of medical data for decision support and prediction. Recent research has demonstrated that artificial intelligence and machine learning is being used to combat the COVID-19 infection. Prediction models can be incorporated and thus aid in designing better strategies and making effective decisions. These technologies analyze past events to make better predictions about what will happen in the future, which may aid in preparation for potential threats and consequences. This survey paper aimed to cover a group of research that uses artificial intelligence applications to predict COVID-19 disease. This survey systematically presented the research to extract data mainly related to the type of article, publication date, research objectives, study context, results, methodology, algorithm, and data set © 2022 IEEE.

Collaborative Attention Transformer on facial expression recognition under partial occlusion

Recent research on facial expression recognition (FER) in the wild shows challenges still remain. Different from laboratory-controlled expression in the past, images in the wild contain more uncertainties, such as different forms of face information occlusion, ambiguous facial images, noisy labels, and so on. Among them, real-world facial occlusion is the most general and crucial challenge for FER. In addition, because of the COVID-19 disease epidemic, people have to wear masks in public, which brings new challenges to FER tasks. Due to the recent success of the Transformer on numerous computer vision tasks, we propose a Collaborative Attention Transformer (CAT) network that first uses Cross-Shaped Window Transformer as the backbone for the FER task. Meanwhile, two attention modules are collaborated. Channel-Spatial Attention Module is designed to increase the attention of the network to global features. Moreover, Window Attention Gate is used to enhance the ability of the model to focus on local details. The proposed method is evaluated on two public in-The-wild facial expression datasets, RAF-DB and FERPlus, and the results demonstrate that our CAT performs superior to the state-of-The-Art methods. © 2022 SPIE and IST.

Deep Learning Analysis for COVID-19 Using Neural Network Algorithms

The COVID-19 pandemic threatens to devastatingly impact the global population’s safety. A successful surveillance of contaminated patients is a crucial move in the battle against COVID-19, and radiological photographs via chest X-ray are one of the main screening strategies. Recent research showed that patients have abnormalities in photographs of chest X-ray that are characteristic of COVID-19 infects. This has inspired a set of deep learning artificial intelligence (AI) programs, and it has been seen that the precision of the identification of COVID-19 contaminated patients utilizing chest X-rays has been quite positive. However, these built AI schemes, to the extent of their author’s awareness, have become closed sources and not accessible for further learning and expansion by the scientific community, so they are not open to the general public. This thesis therefore implements COVID-Net to identify COVID-19 cases of chest X-rays images, an open source, accessible to the general public, a deep neural network architecture adapted to the detection. The COVID-Net data collection, which is referred to as COVIDx which includes 13,800 chest X-ray photographs of 13,725 patients from 3 open-access data sources, one of which we launched, are also addressed. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

A Low-cost Materials Laboratory Sequence for Remote Instruction that Supports Student Agency

Under the new ABET accreditation framework, students are expected to demonstrate “an ability to develop and conduct appropriate experimentation, analyze and interpret data, and use engineering judgment to draw conclusions” [1]. Traditional, recipe-based labs provide few opportunities for students to engage in realistic experimental design, and recent research has cast doubt on their pedagogical benefit [2]. At the same time, the COVID-19 pandemic has forced institutions to move to remote learning. To address these challenges we developed a series of online labs for an upper-division mechanics of materials course. The first three labs consist of video demonstrations of traditional lab experiments with synchronous group discussions and data analysis. Two of these “traditional” virtual labs are supplemented with peer-teaching video activities. The final lab is a guided-inquiry activity focused on experimental design. Using only materials available at home, students measure the Young's modulus of aluminum and use their results to design a hypothetical product. In order to provide the same opportunity for students around the world, the test specimen is taken from an aluminum beverage can. One measure of whether or not an activity supports student agency is the diversity of solutions generated by students [3]. We analyzed 36 reports from the final guided-inquiry lab and coded the experimental procedure on five key decisions such as the type of experiment performed, specimen geometry, and measurement method. We identified 29 unique approaches to the problem, with no one approach accounting for more than three submissions. Student outcomes were measured by a survey of students' attitudes and self-efficacy administered directly after every lab activity except for the first one. The fraction of students endorsing statements related to a sense of agency increased dramatically between the “traditional” labs and the guided-inquiry lab: from 52% to 82% for goal-setting and from about 64% to 92% for choice of methods. Self-efficacy increased significantly in the primary targeted skills (designing experiments, making predictions, and generating further questions), but there was no significant shift in skills not explicitly targeted by the guided-inquiry lab (equitable sharing of labor, expressing opinions in a group, and interpreting graphs). Our experience demonstrates that at-home lab activities can achieve sophisticated learning outcomes without the use of lab equipment or customized kits. © American Society for Engineering Education, 2021