Search

As electronics and the increasing popularity of social media are widely used, a large amount of text data is created at unprecedented rates. All data created cannot be read by humans, and what they discuss in their sphere of interest may be found. Modeling of themes is a way to identify subjects in a vast number of texts. There has been a lot of study on subject-modeling in English. At the same time, millions of people worldwide speak Tamil; there is no great development in resource-scarce languages such as Tamil being spoken by millions of people worldwide. The consequences of specific deep learning models are usually difficult to interpret for the typical user. They are utilizing various visualization techniques to represent the outcomes of deep learning in a meaningful way. Then, they use metrics like similarity, correlation, perplexity, and coherence to evaluate the deep learning models.

Skewed distributions appear in many real-world classification problems. Skewed distributions, underrepresented classes, and multiple overlapping regions in multiclass imbalanced datasets deteriorate the performance of existing classification algorithms and approaches. In this context, we combine a novel preprocessing procedure to tackle minority classes in multiclass imbalanced problems with an ensemble framework. The preprocessing method oversamples the minority classes based on normalized probability, and then an ensemble called a stacked generalization framework is used to train the model. The motive behind combining the ensemble framework and the preprocessing procedure is to enhance the overall classification performance of the classifier for multiclass imbalanced problems. Experimental results on 20 multiclass imbalanced datasets show that the proposed preprocessing method with the ensemble framework outperforms the representative approaches in 13 datasets for macro average arithmetic (MAvA) and mean F-measure (MFM) metrics. In the case of state-of-the-art techniques, the proposed approach steered 14 datasets for the MAvA metric and 15 datasets for the MFM metric to success.

The problem of data skewness, class imbalance, data scarcity and noise limits the application of machine learning and deep learning models in applications like anomaly detection, fraud detection, intrusion detection, fault diagnosis, machine-to-machine communication, etc. Performance of supervised learning leans towards majority class and fails to generalize testing data in class imbalance and noisy data problems. Using neural-based data augmentation techniques for data generation and deep convolutional models for classification would enhance the performance of the applications mentioned above. Recently, GANs (generative adversarial networks) showed significant improvements in generating images. In this paper, a model that uses a conditioned deep convolutional GAN and an auxiliary classifier are proposed to tackle the aforementioned issues. The conditioned GAN is used for data generation of minority classes images and noisy images. Another auxiliary deep convolutional model is employed for the classification of images on data augmented dataset. Also, a multi-hinge loss is employed in both the data augmentation and classification tasks. The effectiveness of the proposed model is investigated on the quality of generated images and classification metrics using four publicly available popular datasets: MNIST, EMNIST, CIFAR10, and SVHN. The proposed model has shown significant improvements over the other often used models of data augmentation and multi-class imbalance image classification in terms of generated samples and classification accuracy.

A global pandemic has reduced mobility, impacting the healthcare assessment, and diagnostics on co-morbid patients. Various challenges for location finding and indoor signalling issues give rise to the necessity for compact and low-cost ultra-wideband sensor technology. Lockdown has played a vital role during crisis in engaging with specific conditions like position estimation in areas under LOS and NLOS caused by diffraction and reflection of UWB signals across obstacles. UWB-based localisation system with edge/fog computing is used to analyse the indoor/outdoor usage of quarantined patients reducing the transmission, which becomes a threat for society and healthcare personals. The mathematical models to symbolise various patterns utilising UWB sensors with anchors and tags were used to determine the flow of all rapid and random movements within the specified clustered locations. The proposed system achieves high classification accuracy with SVM to classify the patterns and assist frontline medical workers in containing pandemics through human interaction.

Location-based identification of entities determines to provide support at the point of need and the precise position in real-time. The indoor parking environments are challenging with complex situations depending on the size, location, and structure with specified locations allocated for entry and exit. UWB provides a high precision positioning for a specific layout through the tags and anchors with accuracy at the centimetre level. In this paper, a specified indoor parking layout is identified with graph entities by visualising the trajectory through SimulNet. The indoor positioning system utilises the NLOS/LOS signal to classify whether the parking lot slots are occupied or not. The UWB positioning mode allows for defining the spatial positioning in the indoor parking lot layout. Decluttered parking spaces with reduced waiting time through graph edge node detection with specific paths for entry are validated through the UWB system. The proposed algorithm provides a real-time dynamic tracking of empty and full parking space with high accuracy.