The identification of hate and offensive content in multilingual social media posts remains challenging due to the high cost of manual annotation and the complexity of code-mixed language. Most existing methods depend on large amounts of fully labeled data, which is not practical for real-world moderation systems. In this work, we propose a BERT-based active learning framework that reduces the annotation requirement to only 25

Canine mammary tumours (CMTs) represent a valuable spontaneous model for human breast cancer, as both share comparable pathological features and frequent alterations in the TP53 pathway. Although TP53 involvement in tumour development is well established, the diagnostic relevance of circulating TP53 autoantibodies in dogs remains insufficiently characterized. This study evaluated the diagnostic utility of TP53 autoantibodies in CMTs using a recombinant TP53-based immunoassay, supported by comparative analyses in human and experimental animal models. Recombinant canine TP53 protein was expressed and employed in indirect ELISA and dot blot assays to detect serum TP53 autoantibodies in 141 canine serum samples, including sera from dogs with mammary tumours and healthy controls. Comparative evaluation included serum samples from 100 human subjects (50 breast cancer patients and 50 healthy individuals) as well as a longitudinal N-nitroso-N-methylurea (NMU)–induced rat mammary tumour model to examine the temporal appearance of autoantibodies before and during tumour development.TP53 autoantibodies were detected in 60.0

The growing influence of social media has reshaped how disaster information is shared in real time. Consequently, rapid identification of actionable content from these platforms is critical for situational awareness and effective humanitarian response. We propose an explainable hybrid model that fuses DeBERTa-v3 contextual embeddings with a compact set of handcrafted linguistic and domain cues (length, lexical richness, punctuation/emphasis, metadata markers, and earthquake-specific keywords, etc). Handcrafted features are z-score standardised, projected to a dense space, then concatenated with the [CLS] embedding for classification. On the HumAID Earthquake subset, the hybrid framework achieves a weighted-average F1-Score of 0.85, outperforming a DeBERTa + LIME baseline at 0.83 (+2.0 points), alongside gains in accuracy and per-class ROC-AUC, especially for minority humanitarian categories. These results show that enriching transformer representations with lightweight, interpretable features yields measurable performance gains and actionable explanations for trustworthy crisis informatics.

Glaucoma is a leading cause of irreversible blindness worldwide, making early and reliable detection essential for effective clinical management. This study proposes a deep hybrid ensemble framework for automated glaucoma detection from retinal fundus images by integrating convolutional and transformer-based architectures. The framework combines EfficientNetB4, Vision Transformer (ViT), and Bidirectional Encoder from Image Transformers (BEiT) to jointly capture local retinal structures and global contextual representations. A feature-level attention-based fusion mechanism is employed to effectively integrate complementary features from these models. Experimental evaluation on the G1020 dataset shows that the proposed framework outperforms individual models and conventional ensemble methods, achieving an accuracy of 98.59%, sensitivity of 96.51%, specificity of 99.45%, and an AUC of 0.9989. To assess generalization capability, the trained model was further validated on the external ACRIMA dataset, where it achieved 96.37% accuracy and an AUC of 0.9881. These results demonstrate that the proposed hybrid architecture provides robust and reliable glaucoma detection across datasets. The framework has the potential to support AI-assisted glaucoma screening and future digital twin-based precision ophthalmology systems.

In times of natural disasters and crises, social media sites are crucial sources of real-time information. Yet, the sheer volume and multimodal nature of this data—encompassing both text and images—make it difficult to quickly identify relevant and actionable content. This study presents a multimodal deep learning model that integrates Bidirectional Encoder Representations from Transformers (BERT) for text and Vision Transformer (ViT) for images. A fusion mechanism is used to combine both modalities for more accurate classification of posts as informative or non-informative. The proposed model is evaluated on benchmark crisis datasets with extensive preprocessing, data augmentation, and regularization strategies to ensure robustness across various disaster scenarios. Experimental results demonstrate that the multimodal approach significantly outperforms unimodal models. The BERT-based text-only model achieves an F1 score of 0.82, the ViT-based image-only model reaches 0.87, while the multimodal BERT+ViT model achieves the highest F1 score of 0.89. Furthermore, it attains an AUC of 0.97 and a validation accuracy of 0.91, confirming the advantage of integrating textual and visual features for real-time disaster informativeness detection. These results highlight the model’s potential to assist emergency responders in filtering high-priority content during critical situations.

Cattle identification through biometrics has increasingly relied on non-invasive methods, where facial and muzzle images are commonly used. While facial images alone offer valuable features, they can be sensitive to variations in lighting, pose, and occlusion, leading to inconsistent results. Similarly, relying solely on muzzle images, though distinct, may not provide sufficient discriminatory power for accurate identification in all cases. Combining both facial and muzzle features offers a more comprehensive solution, overcoming the limitations of using either feature independently. Recent advancements in deep neural networks have significantly influenced this field, but achieving an optimal balance between model accuracy and computational efficiency remains a challenge. In this paper, we present a novel approach by introducing a Cross-Attention Encoder in combination with a Pairwise Triplet Loss function for cattle verification. This method processes face and muzzle images in parallel, enhancing the integration of features from both inputs. The cross-attention mechanism enables the model to focus on the most relevant regions, improving feature alignment and discrimination. With a compact architecture of only 0.6 million parameters, the encoder effectively captures essential features from both the face and muzzle, ensuring precise cattle verification without excessive computational demands. Our approach achieved a testing accuracy of 93.67%, with an average inference time of 35ms per sample, demonstrating the model's efficiency. These findings highlight the strength of our attention-based method in delivering high accuracy and computational performance for cattle verification tasks.

Social media hate speech is a significant issue because it may incite violence, discrimination, and social unrest. Anonymity and reach of such platforms enable the rapid spread of harmful content, targeting individuals or communities based on race, gender, religion, or other attributes. The detection of hate speech is very important for the creation of safe online environments, protection of marginalized groups, and compliance with legal and ethical standards. This paper aims to analyze complex social media content using a combination of textual and audio features. The experimental results establish the effectiveness of the proposed approach, with F1-scores reaching 72% for Tamil, 77% for Malayalam, and 36% for Telugu. Such results strongly indicate that multimodal methodologies have significant room for improvement in hate speech detection in resource-constrained languages and underscore the need to continue further research into this critical area.

The rapid rise of social media has emphasized the spread of harmful and hateful content, making it challenging for its identification. Contextual semantics is very important as prior studies present that context level semantics is a more trustworthy indicator of hatefulness than word level semantics for detecting hate speech. This paper attempts to check the usability of transformer-based models for the identification of hate speech on code-mixed datasets, which includes Google-MuRIL, LaBSE, XLM-Roberta-base, mbert and distil-mbert. The above is largely due to its ability for high-level representations of complex and context-dense meaning. Besides this, we experiment on ensemble approach that covers all of the above models to reach out for an even higher level of performance in detection. The experiment results show the best performing macro F1-scores are reported in case of MuRIL in comparison to other implemented models.

Ensuring reliability and interpretability in AI-driven medical imaging is vital for fostering trust in healthcare applications. Deep learning models, including Convolutional Neural Networks (CNNs), attention-based Transformers, and Mamba-based state-space models, have demonstrated significant advancements in classification performance. However, their black-box nature necessitates rigorous explainability analysis to ensure transparency and reliability. In this study, we systematically evaluate the classification performance of these architectures across three medical imaging datasets: BreakHis for histopathology (at multiple magnifications), Chest X-ray for pulmonary disease classification, and Retinal images for ocular disease diagnosis. We employ gradient-based and attention-based post-hoc explainability techniques, including Grad-CAM, Grad-CAM++, and attention rollout mechanisms, to interpret model decisions and visualize feature attributions. The results reveal that CNNs, particularly MobileNetv2 and ResNet50, excel in datasets requiring fine-grained feature extraction, whereas Transformers demonstrate superior performance in tasks emphasizing global dependencies, such as Chest X-ray analysis. Mamba-based models, such as EfficientViM, provide a balance between computational efficiency and interpretability, effectively capturing long-range dependencies in complex datasets like Retinal images. By analyzing decision-making patterns and feature attribution maps, we highlight the trade-offs between classification accuracy, computational efficiency, and model interpretability. Our findings offer actionable insights for selecting task-specific AI architectures, ensuring a balance between performance and transparency, and paving the way for the deployment of trustworthy AI in medical diagnostics.

Retrieving actionable information from disaster-related social media tweets is critical for disaster response. This study proposes a hybrid framework with post-hoc explainability that associates the DeBERTa-v3-base encoder with a neuro-fuzzy (ANFIS-inspired) reasoning layer for earthquake tweet classification. We fine-tuned and evaluated the model on the dataset: HumAID Earthquake subset, comprising 6,250 training and 1,773 testing tweets across nine humanitarian class categories. SHAP (Shapley Additive explanations) produces local and global token-level importance scores to ensure explainability, yielding classspecific rationales. The proposed model achieves a weighted F1-score of 0.85, surpassing a strong DeBERTa-v3-base baseline (0.83; +2.0 points improvement) with minimal added parameters. Per-class ROC-AUC and confusion matrix analyses highlight systematic error patterns and enhance the model’s transparency. Overall, the framework delivers an accurate, lightweight, and interpretable solution suitable for real-time integration into disaster management pipelines.

Advancements in medical imaging and deep learning have enabled the development of intelligent systems that assist clinicians in diagnosing complex pulmonary diseases. This study addresses the growing concern over lung abnormalities caused by diseases such as COVID-19, tuberculosis (TB), and pneumonia. We propose a convolutional neural network (CNN)-based multi-class classification framework that uses chest X-ray images to automatically detect COVID-19, TB, pneumonia, and normal conditions. The original publicly available dataset exhibited class imbalance, with significantly fewer COVID-19 cases compared to other categories. To address this, the Synthetic Minority Oversampling Technique (SMOTE) are applied at the feature level, generating a balanced dataset of 6,000 chest X-ray images equally distributed across the four classes. The preprocessing techniques have been used to enhance model generalisation, including image normalization, augmentation, and resizing. We evaluated multiple deep learning architectures, including ResNet-50, EfficientNet, DenseNet, and VGG-19. Among these, VGG-19 achieved the highest test accuracy of 97.5%, with precision, recall, and F1-score all exceeding 96% across classes. This unified deep learning pipeline integrates data preprocessing, feature extraction, and classification. The proposed model is intended as a research framework and is currently non-clinical; however, it demonstrates promising potential and could be further explored for assisting radiologists in diagnostic decision-making.

Social networks strive to offer positive content to users, yet a considerable amount of inappropriate material, such as rumors, fake news, and hate speech, persists. Despite significant efforts to detect and prevent hate speech early, it remains widespread due to issues like misspellings and mixed language in posts. To address these challenges, this research utilizes advanced algorithms like CNN, LSTM, and BERT to develop an automated system for detecting hate speech in Telugu-English code-mixed posts. Additionally, evaluating the effectiveness of data translation and transliteration approaches for detecting hate in mixed language. Results indicate that the transliteration approach achieves the highest accuracy, with a performance of 75% accuracy, surpassing raw and translated data by 1% and 3%, respectively. The proposed system may effectively minimizes hate speech and offensive content on social media platforms, resulting in an enhanced user experience. From a managerial perspective, this research presents numerous benefits, such as improved content moderation, optimized resource allocation, data-driven decision-making, enhanced user satisfaction, strengthened reputation management, and greater scalability. These advancements underscore the potential of utilizing advanced technologies to address complex challenges in social media management.

Cloud computing is one of the foundation technologies of Industry 4.0. Cloud 2.0 is the upcoming cloud technology that addresses several bottlenecks of Cloud 1.0. For instance, the presence of small service providers is threatened by the dominance of a few giant service providers in today's cloud market in Cloud 1.0. Under this circumstance, the small service providers must work together to compete with the giant competitors to survive in the market. For that, small service providers require a transparent, fair, cost-effective, fault-tolerant, and easily scalable platform that can provide reliable and quality services to customers. This work introduces a blockchain-based framework to provide such a platform for cloud service providers and their customers. Here, a new consensus mechanism is proposed to maintain the system's fairness, decentralization, and consistency. A consensus-based service monitoring concept is also introduced to assess the service quality. If a service provider does not deliver the committed quality of service (QoS), a penalty is imposed on the service provider. This framework is designed so that the service providers are always bound to provide committed QoS to the customers. Finally, we performed several experiments, and the experimental results corroborate our claims regarding the proposed framework.

Skin lesions encompass various skin conditions, including cancerous growths resulting from uncontrolled proliferation of skin cells. Globally, this disease affects a significant portion of the population, with millions of fatalities recorded. Over the past three decades, there has been a concerning escalation in diagnosed cases of skin cancer. Early detection is crucial for effective treatment, as late diagnosis significantly heightens mortality risk. Existing research often focuses on either handcrafted or deep features, neglecting the diverse textural and structural properties inherent in skin lesion images. Additionally, reliance on a single optimizer in CNN- based schemes poses efficiency challenges. To tackle these issues, this paper presents two novel approaches for classifying skin lesions in dermoscopic images to assess cancer severity. The first approach enhances classification accuracy by leveraging a modified VGG-16 network and employing both RMSProp and Adam optimizers. The second approach introduces a Hybrid CNN Model, integrating deep features from the modified VGG-16 network with handcrafted color and multi-directional texture features. Color features are extracted using a non-uniform cumulative probability-based histogram method, while texture features are derived from a 45 degrees degrees rotated complex wavelet filter-based dual-tree complex wavelet transform. The amalgamated features facilitate accurate prediction of skin lesion classes. Evaluation on ISIC 2017 skin cancer classification challenge images demonstrates significant performance enhancements over existing techniques.

Sarcasm in social media postings significantly impacts automated sentiment extraction due to its potential to invert the overall polarity of phrases. It poses a formidable challenge in extracting genuine sentiments from informal multilingual comments about events or individuals, especially when the message is in a multilingual form. The prevalence of noisy datasets further complicates this task. This work proposes a hybrid convolutional neural network (CNN-H) model to classify sarcastic and non-sarcastic statements. Unlike traditional CNN models, CNN-H integrates character and word embeddings, enhancing its capacity to handle the complexities of multilingual content. Evaluation on five benchmark datasets, including English (news headlines, Ghosh Tweet, Riloff Tweet), Hindi, and multilingual Hindi-English (Hinglish), demonstrates the superior performance of CNN-H. The model surpasses various state-of-the-art conventional machine-learning and deep-learning models, including CNN, achieving an impressive F1-score of 93.0