<![CDATA[Convolutional Neural Network is one of the Deep Learning Algorithms. CNN itself is developed from Multilayer Peceptron (MLP) method. CNN and MLP are algorithms that focused on processing data in two dimensions form, such as pictures or sounds. CNN is made with the principle of translation invariance. That means CNN is able to recognize objects at various possible positions. There are 150 sign language images that are classified using Alexnet in this system. Alexnet is Krizhevsky's work at developing CNN method as a clessifier. CNN architecture developed by Alex has eight feature extraction layers. The layer consists of five convolution layers and three pooling layers. In its classification layer, Alexnet has two fully connected layers, each of them has 4096 neurons. At the end of the layer, there are classifications into 5 categories using softmax activation. The average accuracy of the classification results even reaches 100%.]]>
Copyrights © 2020
