Speaking or expressing oneself in an abusive manner is a form of verbal abuse that targets individuals or groups on the basis of their membership in a particular social group, which is differentiated by traits such as culture, gender, sexual orientation, religious affiliation etc. In today's world, the dissemination of evil and depraved content on social media has increased exponentially. Abusive language on the internet has been linked to an increase in violence against minorities around the world, including mass shootings, murders, and ethnic cleansing. People who use social media in places where English is not the main language often use a code-mixed form of text. This makes it harder to find abusive texts, and when combined with the fact that there aren't many resources for languages like Tamil, the task becomes significantly challenging. This work makes use of abusive Tamil language comments released by the workshop “Tamil DravidianLangTech@ACL 2022” and develops adapter-based multilingual transformer models namely Muril, XLMRoBERTa and mBERT to classify the abusive comments. These transformers have been utilized as fine-tuners and adapters. This study shows that in low-resource languages like Tamil, adapter-based strategies work better than fine-tuned models. In addition, we use Optuna, a hyperparameter optimization framework to find the ideal values of the hyper-parameters that lead to better classification. Of all the proposed models, MuRIL (Large) gives 74.7%, which is comparatively better than other models proposed for the same dataset.

As COVID-19 spreads rapidly all over the world, the lack of reliable testing kits and medical diagnoses makes the infection more vulnerable to the human population. An effective diagnosis and detection of the SARS-Cov-2 virus are required to control and prevent the COVID-19 disease. In this study, we employed a convolution neural network (CNN) to detect coronavirus-infected patients using computed tomography (CT) images. The proposed study utilized transfer learning on the three pre-trained deep CNN models to detect COVID-19 infection from the chest CT scan images. We have tuned and optimized the hyper-parameters of the pre-trained CNN models using the Bayesian Optimization technique. Further, the deep CNN architectures are incorporated with the Learning without Forgetting (LwF) technique to improve the model’s capability to recognize new Delta variants COVID-19 data. The CNN model with the LwF is evaluated on the CT images of original and the Delta-variant COVID-19 dataset. The performance of the learning, without forgetting based CNN models namely VGG16, InceptionV3, and Xception is assessed using different performance evaluation metrics in detecting COVID-19 disease. The experimental result shows that the Xception model’s performance is superior that other two developed models and effective in classifying original augmented images and new Delta-variant images with an accuracy of 98.31% and 92.32%, respectively.The empirical result shows our model performance is significantly effective in diagnosis and classification of two different variants of the SARS-CoV-2 virus and the developed CNN models can provide assistance to the medical experts for diagnosing different variants of COVID-19 disease.