scholarly journals Offensive-Language Detection on Multi-Semantic Fusion Based on Data Augmentation

2022 ◽  
Vol 5 (1) ◽  
pp. 9
Author(s):  
Junjie Liu ◽  
Yong Yang ◽  
Xiaochao Fan ◽  
Ge Ren ◽  
Liang Yang ◽  
...  
Keyword(s):  
Data Augmentation ◽  
Rapid Identification ◽  
Semantic Features ◽  
Fusion Model ◽  
Viral Spread ◽  
Word Level ◽  
Back Translation ◽  
Public Datasets ◽  
The Impact ◽  
Offensive Language

The rapid identification of offensive language in social media is of great significance for preventing viral spread and reducing the spread of malicious information, such as cyberbullying and content related to self-harm. In existing research, the public datasets of offensive language are small; the label quality is uneven; and the performance of the pre-trained models is not satisfactory. To overcome these problems, we proposed a multi-semantic fusion model based on data augmentation (MSF). Data augmentation was carried out by back translation so that it reduced the impact of too-small datasets on performance. At the same time, we used a novel fusion mechanism that combines word-level semantic features and n-grams character features. The experimental results on the two datasets showed that the model proposed in this study can effectively extract the semantic information of offensive language and achieve state-of-the-art performance on both datasets.

scholarly journals Data augmentation for computed tomography angiography via synthetic image generation and neural domain adaptation

2020 ◽  
Vol 6 (1) ◽  
Author(s):  
Malte Seemann ◽  
Lennart Bargsten ◽  
Alexander Schlaefer
Keyword(s):  
Computed Tomography ◽  
Neural Networks ◽  
Deep Learning ◽  
Medical Imaging ◽  
Computed Tomography Angiography ◽  
Data Augmentation ◽  
Domain Adaptation ◽  
Synthetic Image ◽  
Wide Range ◽  
The Impact

AbstractDeep learning methods produce promising results when applied to a wide range of medical imaging tasks, including segmentation of artery lumen in computed tomography angiography (CTA) data. However, to perform sufficiently, neural networks have to be trained on large amounts of high quality annotated data. In the realm of medical imaging, annotations are not only quite scarce but also often not entirely reliable. To tackle both challenges, we developed a two-step approach for generating realistic synthetic CTA data for the purpose of data augmentation. In the first step moderately realistic images are generated in a purely numerical fashion. In the second step these images are improved by applying neural domain adaptation. We evaluated the impact of synthetic data on lumen segmentation via convolutional neural networks (CNNs) by comparing resulting performances. Improvements of up to 5% in terms of Dice coefficient and 20% for Hausdorff distance represent a proof of concept that the proposed augmentation procedure can be used to enhance deep learning-based segmentation for artery lumen in CTA images.

scholarly journals Riding the waves: the ongoing impact of COVID-19 on a national surgical training cohort

2021 ◽  
Author(s):  
Orla Hennessy ◽  
Amy Lee Fowler ◽  
Conor Hennessy ◽  
David Brinkman ◽  
Aisling Hogan ◽  
...  
Keyword(s):  
Surgical Training ◽  
World Health Organisation ◽  
World Health ◽  
Training Methods ◽  
Viral Spread ◽  
Surgical Services ◽  
Global Pandemic ◽  
Surgical Trainees ◽  
Secondary Survey ◽  
The Impact

Abstract Background The World Health Organisation declared a global pandemic on the 11 March 2020 resulting in implementation of methods to contain viral spread, including curtailment of all elective and non-emergent interventions. Many institutions have experienced changes in rostering practices and redeployment of trainees to non-surgical services. Examinations, study days, courses, and conferences have been cancelled. These changes have the potential to significantly impact the education and training of surgical trainees. Aim To investigate the impact of the COVID-19 pandemic on training, educational, and operative experiences of Irish surgical trainees. Methods Surgical trainees were surveyed anonymously regarding changes in working and educational practices since the declaration of the COVID-19 pandemic on 11 March 2020. The survey was circulated in May 2020 to both core and higher RCSI surgical trainees, when restrictions were at level five. Questions included previous and current access to operative sessions as well as operative cases, previous and current educational activities, access to senior-led training, and access to simulation-/practical-based training methods. A repeat survey was carried out in October 2020 when restrictions were at level two. Results Overall, primary and secondary survey response rates were 29% (n = 98/340) and 19.1% (n = 65/340), respectively. At the time of circulation of the second survey, the number of operative sessions attended and cases performed had significantly improved to numbers experienced pre-pandemic (p < 0.0001). Exposure to formal teaching and education sessions returned to pre-COVID levels (p < 0.0001). Initially, 23% of trainees had an examination cancelled; 53% of these trainees have subsequently sat these examinations. Of note 27.7% had courses cancelled, and 97% of these had not been rescheduled. Conclusion Surgical training and education have been significantly impacted in light of COVID-19. This is likely to continue to fluctuate in line with subsequent waves. Significant efforts have to be made to enable trainees to meet educational and operative targets.

scholarly journals Adversarial Data Augmentation on Breast MRI Segmentation

2021 ◽  
Vol 11 (10) ◽  
pp. 4554
Author(s):  
João F. Teixeira ◽  
Mariana Dias ◽  
Eva Batista ◽  
Joana Costa ◽  
Luís F. Teixeira ◽  
...  
Keyword(s):  
Data Augmentation ◽  
Medical Image Analysis ◽  
Synthetic Data ◽  
Breast Mri ◽  
Semantic Segmentation ◽  
Adversarial Networks ◽  
Augmentation Strategy ◽  
Magnetic Resonance Imaging Mri ◽  
Segmentation Task ◽  
The Impact

The scarcity of balanced and annotated datasets has been a recurring problem in medical image analysis. Several researchers have tried to fill this gap employing dataset synthesis with adversarial networks (GANs). Breast magnetic resonance imaging (MRI) provides complex, texture-rich medical images, with the same annotation shortage issues, for which, to the best of our knowledge, no previous work tried synthesizing data. Within this context, our work addresses the problem of synthesizing breast MRI images from corresponding annotations and evaluate the impact of this data augmentation strategy on a semantic segmentation task. We explored variations of image-to-image translation using conditional GANs, namely fitting the generator’s architecture with residual blocks and experimenting with cycle consistency approaches. We studied the impact of these changes on visual verisimilarity and how an U-Net segmentation model is affected by the usage of synthetic data. We achieved sufficiently realistic-looking breast MRI images and maintained a stable segmentation score even when completely replacing the dataset with the synthetic set. Our results were promising, especially when concerning to Pix2PixHD and Residual CycleGAN architectures.

scholarly journals A Qualitative Study of COVID-19 Distress in University Students

2021 ◽  
pp. 216769682110251
Author(s):  
Samantha G. Farris ◽  
Mindy M. Kibbey ◽  
Erick J. Fedorenko ◽  
Angelo M. DiBello
Keyword(s):  
University Students ◽  
Undergraduate Students ◽  
Mental Health Problems ◽  
Financial Strain ◽  
Public Response ◽  
Viral Spread ◽  
Future Work ◽  
The Impact ◽  
Viral Outbreak ◽  
Selection Of

The psychological effect of the pandemic and measures taken in response to control viral spread are not yet well understood in university students; in-depth qualitative analysis can provide nuanced information about the young adult distress experience. Undergraduate students ( N = 624) in an early US outbreak “hotspot” completed an online narrative writing about the impact and distress experienced due to the COVID-19 pandemic. Data were collected April-May 2020. A random selection of 50 cases were sampled for thematic analysis. Nine themes were identified: viral outbreak distress, fear of virus contraction/transmission, proximity to virus, dissatisfaction with public response, physical distancing distress, social distancing distress, academic and school-related distress, disruptive changes in health behavior and routines, financial strain and unemployment, worsening of pre-existing mental health problems, and social referencing that minimizes distress. Future work is needed to understand the persistence of the distress, in addition to developing methods for assessment, monitoring, and mitigation of the distress.

scholarly journals Offensive keyword extraction based on the attention mechanism of BERT and the eigenvector centrality using a graph representation

2021 ◽  
Author(s):  
Gretel Liz De la Peña Sarracén ◽  
Paolo Rosso
Keyword(s):  
Language Model ◽  
Hybrid Approach ◽  
Attention Mechanism ◽  
Graph Representation ◽  
Keyword Extraction ◽  
Eigenvector Centrality ◽  
Model Learning ◽  
Harmful Content ◽  
The Impact ◽  
Offensive Language

AbstractThe proliferation of harmful content on social media affects a large part of the user community. Therefore, several approaches have emerged to control this phenomenon automatically. However, this is still a quite challenging task. In this paper, we explore the offensive language as a particular case of harmful content and focus our study in the analysis of keywords in available datasets composed of offensive tweets. Thus, we aim to identify relevant words in those datasets and analyze how they can affect model learning. For keyword extraction, we propose an unsupervised hybrid approach which combines the multi-head self-attention of BERT and a reasoning on a word graph. The attention mechanism allows to capture relationships among words in a context, while a language model is learned. Then, the relationships are used to generate a graph from what we identify the most relevant words by using the eigenvector centrality. Experiments were performed by means of two mechanisms. On the one hand, we used an information retrieval system to evaluate the impact of the keywords in recovering offensive tweets from a dataset. On the other hand, we evaluated a keyword-based model for offensive language detection. Results highlight some points to consider when training models with available datasets.

scholarly journals The Impact on Infectivity and Neutralization Efficiency of SARS-CoV-2 Lineage B.1.351 Pseudovirus

Viruses ◽  
2021 ◽  
Vol 13 (4) ◽  
pp. 633
Author(s):  
Yeong Jun Kim ◽  
Ui Soon Jang ◽  
Sandrine M. Soh ◽  
Joo-Youn Lee ◽  
Hye-Ra Lee
Keyword(s):  
South Africa ◽  
Monoclonal Antibodies ◽  
Cell Fusion ◽  
Receptor Binding ◽  
Neutralizing Antibodies ◽  
Receptor Binding Domain ◽  
Viral Spread ◽  
New Variant ◽  
Global Concern ◽  
The Impact

A new variant of SARS-CoV-2 B.1.351 lineage (first found in South Africa) has been raising global concern due to its harboring of multiple mutations in the spike that potentially increase transmissibility and yield resistance to neutralizing antibodies. We here tested infectivity and neutralization efficiency of SARS-CoV-2 spike pseudoviruses bearing particular mutations of the receptor-binding domain (RBD) derived either from the Wuhan strains (referred to as D614G or with other sites) or the B.1.351 lineage (referred to as N501Y, K417N, and E484K). The three different pseudoviruses B.1.351 lineage related significantly increased infectivity compared with other mutants that indicated Wuhan strains. Interestingly, K417N and E484K mutations dramatically enhanced cell–cell fusion than N501Y even though their infectivity were similar, suggesting that K417N and E484K mutations harboring SARS-CoV-2 variant might be more transmissible than N501Y mutation containing SARS-CoV-2 variant. We also investigated the efficacy of two different monoclonal antibodies, Casirivimab and Imdevimab that neutralized SARS-CoV-2, against several kinds of pseudoviruses which indicated Wuhan or B.1.351 lineage. Remarkably, Imdevimab effectively neutralized B.1.351 lineage pseudoviruses containing N501Y, K417N, and E484K mutations, while Casirivimab partially affected them. Overall, our results underscore the importance of B.1.351 lineage SARS-CoV-2 in the viral spread and its implication for antibody efficacy.

scholarly journals Cross-evaluation of social mining for classification of depressed online personas

2021 ◽  
Vol 0 (0) ◽  
Author(s):  
Alina Trifan ◽  
José Luis Oliveira
Keyword(s):  
Transfer Learning ◽  
Early Stage ◽  
Post Partum ◽  
Evaluation Study ◽  
Continuous Increase ◽  
Digital Phenotyping ◽  
Public Datasets ◽  
The Impact ◽  
Social Mining

Abstract With the continuous increase in the use of social networks, social mining is steadily becoming a powerful component of digital phenotyping. In this paper we explore social mining for the classification of self-diagnosed depressed users of Reddit as social network. We conduct a cross evaluation study based on two public datasets in order to understand the impact of transfer learning when the data source is virtually the same. We further complement these results with an experiment of transfer learning in post-partum depression classification, using a corpus we have collected for the matter. Our findings show that transfer learning in social mining might still be at an early stage in computational research and we thoroughly discuss its implications.

scholarly journals Wolbachia-Mediated Antiviral Protection in Drosophila Larvae and Adults following Oral Infection

2015 ◽  
Vol 81 (23) ◽  
pp. 8215-8223 ◽  
Author(s):  
Aleksej L. Stevanovic ◽  
Pieter A. Arnold ◽  
Karyn N. Johnson
Keyword(s):  
Virus Infection ◽  
Oral Delivery ◽  
Life Stage ◽  
Potential Effect ◽  
Oral Exposure ◽  
Viral Dynamics ◽  
Content Type ◽  
Viral Spread ◽  
Infected Insect ◽  
The Impact

ABSTRACTUnderstanding viral dynamics in arthropods is of great importance when designing models to describe how viral spread can influence arthropod populations. The endosymbiotic bacteriumWolbachiaspp., which is present in up to 40% of all insect species, has the ability to alter viral dynamics in bothDrosophilaspp. and mosquitoes, a feature that in mosquitoes may be utilized to limit spread of important arboviruses. To understand the potential effect ofWolbachiaon viral dynamics in nature, it is important to consider the impact of natural routes of virus infection onWolbachiaantiviral effects. Using adultDrosophilastrains, we show here thatDrosophila-Wolbachiaassociations that have previously been shown to confer antiviral protection following systemic viral infection also confer protection against virus-induced mortality following oral exposure to Drosophila C virus in adults. Interestingly, a different pattern was observed when the same fly lines were challenged with the virus when still larvae. Analysis of the fourDrosophila-Wolbachiaassociations that were protective in adults indicated that only the w1118-wMelPop association conferred protection in larvae following oral delivery of the virus. Analysis ofWolbachiadensity using quantitative PCR (qPCR) showed that a highWolbachiadensity was congruent with antiviral protection in both adults and larvae. This study indicates thatWolbachia-mediated protection may vary between larval and adult stages of a givenWolbachia-host combination and that the variations in susceptibility by life stage correspond withWolbachiadensity. The differences in the outcome of virus infection are likely to influence viral dynamics inWolbachia-infected insect populations in nature and could also have important implications for the transmission of arboviruses in mosquito populations.

Underwater Acoustic Target Recognition Based on Generative Adversarial Network Data Augmentation

2021 ◽  
Vol 263 (2) ◽  
pp. 4558-4564
Author(s):  
Minghong Zhang ◽  
Xinwei Luo
Keyword(s):  
Data Augmentation ◽  
Target Recognition ◽  
Training Data ◽  
Small Samples ◽  
Generative Adversarial Network ◽  
Data Set ◽  
Underwater Acoustic ◽  
Adversarial Network ◽  
Acoustic Target ◽  
The Impact

Underwater acoustic target recognition is an important aspect of underwater acoustic research. In recent years, machine learning has been developed continuously, which is widely and effectively applied in underwater acoustic target recognition. In order to acquire good recognition results and reduce the problem of overfitting, Adequate data sets are essential. However, underwater acoustic samples are relatively rare, which has a certain impact on recognition accuracy. In this paper, in addition of the traditional audio data augmentation method, a new method of data augmentation using generative adversarial network is proposed, which uses generator and discriminator to learn the characteristics of underwater acoustic samples, so as to generate reliable underwater acoustic signals to expand the training data set. The expanded data set is input into the deep neural network, and the transfer learning method is applied to further reduce the impact caused by small samples by fixing part of the pre-trained parameters. The experimental results show that the recognition result of this method is better than the general underwater acoustic recognition method, and the effectiveness of this method is verified.

scholarly journals Translation and adaptation of the Child Posttraumatic Cognitions Inventory (cPTCI) to Brazilian Portuguese

2014 ◽  
Vol 36 (2) ◽  
pp. 107-112 ◽  
Author(s):  
Beatriz de Oliveira Meneguelo Lobo ◽  
Alice Einloft Brunnet ◽  
Thiago Loreto Garcia da Silva ◽  
Lafaiete Moreira dos Santos ◽  
Gustavo Gauer ◽  
...  
Keyword(s):  
Concurrent Validity ◽  
Stress Disorder ◽  
Traumatic Event ◽  
Target Population ◽  
Brazilian Portuguese ◽  
Subjective Responses ◽  
Posttraumatic Cognitions ◽  
Brazilian Version ◽  
Back Translation ◽  
The Impact

Introduction: Experiencing a traumatic event is a risk factor for the development of mental illness, especially posttraumatic stress disorder. A child's appraisals of a traumatic event may play a prominent role in the development or maintenance of the disorder. Therefore, subjective responses should be evaluated to understand the impact of a traumatic event on a child's life. This study translated and adapted the Child Posttraumatic Cognitions Inventory (cPTCI) for use in linguistic and cultural contexts in Brazil. Methods: The process included translation, back-translation, language expert evaluation and expert committee's evaluation. Results: Content validity index scores were good for all dimensions after evaluation by two judges and one reformulation. The back-translation of the final version also showed that the cPTCI items in Brazilian Portuguese maintained the same meanings of the original in English. This version was tested in a sample of the target population, and all the items were above the cut-off point (minimum = 3.6; maximum = 4.0). Conclusions: This study was successful in producing a Brazilian version of the cPTCI. Further studies are underway to examine the reliability and the factorial and concurrent validity of cPTCI subscales.

Sign in / Sign up

Export Citation Format

Share Document