Abubakar Inuwa Isa - Academia.edu (original) (raw)
Uploads
Papers by Abubakar Inuwa Isa
ArXiv, 2020
Improving neural machine translation (NMT) models using the back-translations of the monolingual ... more Improving neural machine translation (NMT) models using the back-translations of the monolingual target data (synthetic parallel data) is currently the state-of-the-art approach for training improved translation systems. The quality of the backward system - which is trained on the available parallel data and used for the back-translation - has been shown in many studies to affect the performance of the final NMT model. In low resource conditions, the available parallel data is usually not enough to train a backward model that can produce the qualitative synthetic data needed to train a standard translation model. This work proposes a self-training strategy where the output of the backward model is used to improve the model itself through the forward translation technique. The technique was shown to improve baseline low resource IWSLT'14 English-German and IWSLT'15 English-Vietnamese backward translation models by 11.06 and 1.5 BLEUs respectively. The synthetic data generated...
Journal of Racial and Ethnic Health Disparities
Background A relentless flood of information accompanied the novel coronavirus 2019 (COVID-19) pa... more Background A relentless flood of information accompanied the novel coronavirus 2019 (COVID-19) pandemic. False news, conspiracy theories, and magical cures were shared with the general public at an alarming rate, which may lead to increased anxiety and stress levels and associated debilitating consequences. Objectives To measure the level of COVID-19 information overload (COVIO) and assess the association between COVIO and sociodemographic characteristics among the general public. Methods A cross-sectional online survey was conducted between April and May 2020 using a modified Cancer Information Overload scale. The survey was developed and posted on four social media platforms. The data were only collected from those who consented to participate. COVIO score was classified into high vs. low using the asymmetrical distribution as a guide and conducted a binary logistic regression to examine the factors associated with COVIO. Results A total number of 584 respondents participated in this study. The mean COVIO score of the respondents was 19.4 (± 4.0). Sources and frequency of receiving COVID-19 information were found to be significant predictors of COVIO. Participants who received information via the broadcast media were more likely to have high COVIO than those who received information via the social media (adjusted odds ratio ([aOR],14.599; 95% confidence interval [CI], 1.608-132.559; p = 0.017). Also, participants who received COVID-19 information every minute (aOR, 3.892; 95% CI, 1.124-13.480; p = 0.032) were more likely to have high COVIO than those who received information every week. Conclusion The source of information and the frequency of receiving COVID-19 information were significantly associated with COVIO. The COVID-19 information is often conflicting, leading to confusion and overload of information in the general population. This can have unfavorable effects on the measures taken to control the transmission and management of COVID-19 infection.
ArXiv, 2020
Improving neural machine translation (NMT) models using the back-translations of the monolingual ... more Improving neural machine translation (NMT) models using the back-translations of the monolingual target data (synthetic parallel data) is currently the state-of-the-art approach for training improved translation systems. The quality of the backward system - which is trained on the available parallel data and used for the back-translation - has been shown in many studies to affect the performance of the final NMT model. In low resource conditions, the available parallel data is usually not enough to train a backward model that can produce the qualitative synthetic data needed to train a standard translation model. This work proposes a self-training strategy where the output of the backward model is used to improve the model itself through the forward translation technique. The technique was shown to improve baseline low resource IWSLT'14 English-German and IWSLT'15 English-Vietnamese backward translation models by 11.06 and 1.5 BLEUs respectively. The synthetic data generated...
Journal of Racial and Ethnic Health Disparities
Background A relentless flood of information accompanied the novel coronavirus 2019 (COVID-19) pa... more Background A relentless flood of information accompanied the novel coronavirus 2019 (COVID-19) pandemic. False news, conspiracy theories, and magical cures were shared with the general public at an alarming rate, which may lead to increased anxiety and stress levels and associated debilitating consequences. Objectives To measure the level of COVID-19 information overload (COVIO) and assess the association between COVIO and sociodemographic characteristics among the general public. Methods A cross-sectional online survey was conducted between April and May 2020 using a modified Cancer Information Overload scale. The survey was developed and posted on four social media platforms. The data were only collected from those who consented to participate. COVIO score was classified into high vs. low using the asymmetrical distribution as a guide and conducted a binary logistic regression to examine the factors associated with COVIO. Results A total number of 584 respondents participated in this study. The mean COVIO score of the respondents was 19.4 (± 4.0). Sources and frequency of receiving COVID-19 information were found to be significant predictors of COVIO. Participants who received information via the broadcast media were more likely to have high COVIO than those who received information via the social media (adjusted odds ratio ([aOR],14.599; 95% confidence interval [CI], 1.608-132.559; p = 0.017). Also, participants who received COVID-19 information every minute (aOR, 3.892; 95% CI, 1.124-13.480; p = 0.032) were more likely to have high COVIO than those who received information every week. Conclusion The source of information and the frequency of receiving COVID-19 information were significantly associated with COVIO. The COVID-19 information is often conflicting, leading to confusion and overload of information in the general population. This can have unfavorable effects on the measures taken to control the transmission and management of COVID-19 infection.