Disease Classification Model Using Capsule Networks Based on White Blood Cells

As a substitute for recurrent neural networks (RNNs), Deep Learning (DLs) has shown promise in text processing in the last several years. In this study, we looked at the recently announced capsule networks (CapsNets), which are gaining a lot of interest since they outperform Deep Learning (DLs) in image analysis, and we found that they function well for language categorization and sentiment analysis in some instances. Our experimental findings demonstrate that the suggested well-tuned CapsNets model may serve as a suitable (and, in some cases, superior and more cost-effective) alternative to sentence categorization models based on RNNs and DLs. We ran a battery of tests with randomly shuffled test data to see whether CapsNets could learn the sequence of words sequentially. Overall, our CapsNets model outperforms DL and RNN models in terms of classification performance and resilience to adversarial assaults.