Development of an epilepsy diagnosġs method in neonates byclassifying electroencephalography signals with deep neuralnetworks

The paper considers the problem of detecting neonatal seizures via a transfer learning technique that judiciously reconstructs pretrained deep convolution neural networks (p-DCNN), including alexnet, resnet18, googlenet, densenet and resnet50 in this thesis. These networks are utilized to detect neonatal seizure from multi channel electroencephalography (EEG) signals converted to color images. To handle this problem, it may be directly used a deep neural network (DNN) such as a convolution neural network (CNN) instead of transfer learning based networks. However, a DNN requires too much training data, too much training time and a computer with high performance computational capability. The DNN also has a number of user supplied parameters that must be tuned to obtain desirable classication success. In order to prevent these drawbacks, we proposed a transfer learning technique to solve neonatal seizures detection problem. The results of simulations and the statistical analysis enable us to devise a transfer learning technique that can be employed for seizure detection.