Compare the classification performances of convolutional neural networks and capsule networks on the coswara dataset

Convolutional Neural Networks (CNN) and Capsule Networks (CapsNet) are two commonly used types of neural networks in various fields, including computer vision, natural language processing, speech recognition, and robotics.CNNs are feedforward neural networks designed to process inputs using a sequence of convolutional layers. These layers apply filters to extract features from the input, which generates a set of feature maps. The feature maps are then processed by pooling layers that reduce the spatial dimensionality of the data. Finally, one or more fully connected layers generate the final output. On the other hand, Capsule Networks are a relatively new type of neural network that aims to overcome some of the limitations of CNN, such as their inability to handle spatial relationships between features. Capsule Networks use a hierarchical structure of capsules, which are groups of neurons that represent different object properties. These capsules are connected in a way that allows them to learn spatial relationships between features, leading to a more comprehensive object representation.In this study, we applied both Convolutional Neural Networks and Capsule Networks to the Coswara dataset, which contains audio samples for both healthy individuals and those with COVID-19, and is available online. We compared the classification performance of two networks were compared and the classification performance, such as cross-validation, oversampling, and normalization were evaluated. Our results were shown that the Convolutional Neural Network achieved the best results and was faster than the Capsule Network. Moreover, the CNN was less impacted when preprocessing techniques were not applied.