DIGIT CONVOLUTIONAL NEURAL NETWORK

Convolutional Neural Networks are deep artificial neural networks with high efficiency for classifying images, text and audio. It is responsible for boosting the Computer Vision area, with high employability in autonomous cars, drones and medical diagnostics.

​

This architecture employs convolutional masking and data pooling methods, extracting new information from the data regarding neighborhood and similarity. When it comes to image analysis, the model is capable to identify additional information such as edge and shape of objects.

​

We will continue the previous study, over the mnist digit data, where we applied feedforward neural networks. We will analyze now convolutional neural networks for this problem. At the end you can check our final model working and predicting any number you drawn!