Convolution neural networks

Convolution neural network (CNN) structure is based on feed forward neural network and it is designed on an animal cortex and uses multi-​layered perceptron for this process. In CNN the minimum amount of pre-​processing rectified linear unit activation functions are often used. General applications are image/video recognition, natural language processing, chess etc. Convolution is used to find the features which are similar by using different places of images. It is conducted using learnable filters which are passed through the input data/images. The technique used to increase the dataset and improve CNN accuracy is known as data augmentation. Provided that large acceptable big dataset, CNN increases the exactness of correct classification. Some of the applications of ANN are decoding facial reorganization, document analysing, historic and environment collection, understanding of climate, advertising etc.