Performance assessment of Deep Learning procedures on Malaria dataset
Abstract
Malaria detection is a time-consuming procedure. Only blood sample investigation is the practice which provides the confirmation. Now numerous computational methods have been used to make it faster. The proposed model uses the conception of Convolutional Neural Network (CNN) to lessen the time complexity in identification of Malaria. The prototypical model uses different deep learning algorithms which uses the same dataset to validate the stability. Model uses the two various components of CNN like Sequential and ResNet. ResNet uses more of number of hidden layers rather than sequential. The ResNet model achieved 96.50% accuracy on the training data, 96.78% accuracy on the validation data and 97% accuracy on the testing data. Sequential model on the other hand achieved 98% accuracy on the training data, 96% accuracy on the validation data and 96% accuracy on the testing data. From the initial hypothesis, we get to know that there is no significant difference in the accuracy when we have too many layers.
Moraref