dc.contributor.author | Habib, Gousia | |
dc.contributor.author | Qureshi, Shaima | |
dc.date.accessioned | 2020-07-21T13:23:43Z | |
dc.date.available | 2020-07-21T13:23:43Z | |
dc.date.issued | 2020-07-01 | |
dc.identifier.issn | 2210-142X | |
dc.identifier.uri | https://journal.uob.edu.bh:443/handle/123456789/4019 | |
dc.description.abstract | Accurate biomedical image classification is essential for clinical investigation of different hazardous maladies. Fairly diagnosis of the disease is essential to provide proper treatment and to save precious human lives .Classification methods that support handcrafted features and use artificial neural networks trained with restricted data-set do not have the capacity to viably enhance the precision rate and meet the stipulations for classification of biomedical images End-to-End deep learning machines empowers direct mapping from crude information to the desired output, eliminating the need for handcrafted features. Deep learning has proven as a powerful classification method as evidenced by its success in recent computer vision competitions. Unique deep convolutional neural network (CNN) model for brain tumor classification has been proposed in this paper. The model tested on OASIS MRI data-set and gives an average accuracy of 90:84%. The present model is based on pre- trained vgg-19 model on large Image-Net database. Novel CNN model does not require training from scratch wasting weeks or days, rather uses transfer learning for knowledge distillation. Also in order to further enhance the training acceleration other optimization methods have been used, weights are initialized by Gaussian initialization method followed by ReLu activation function. ADAMS SGD optimization has been used and a drop out algorithm is implemented to get rid from overfitting of the model. The model when implemented on the biomedical image dataset have achieved the highest classification accuracy rate, outperforming all existing techniques with lesser training time. | en_US |
dc.language.iso | en | en_US |
dc.publisher | University of Bahrain | en_US |
dc.rights | Attribution-NonCommercial-NoDerivatives 4.0 International | * |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/4.0/ | * |
dc.subject | CNN, SIFT, HOG, PHT, MRI, OASIS, ADAMS, SGD, CONVONET; ReLu; Soft-max; LRN; FC | en_US |
dc.title | Biomedical Image Classification using CNN by Exploiting Deep Domain Transfer Learning | en_US |
dc.type | Article | en_US |
dc.identifier.doi | http://dx.doi.org/10.12785/ijcds/100197 | |
dc.volume | 10 | en_US |
dc.pagestart | 2 | en_US |
dc.pageend | 11 | en_US |
dc.source.title | International Journal of Computing and Digital Systems | en_US |
dc.abbreviatedsourcetitle | IJCDS | en_US |
The following license files are associated with this item: