Convolutional Neural Networks

Now that we have learned about neural networks and DL, let’s look at some business use cases.

The first case is image recognition. How can we teach a computer to recognize an image? It is an easy task for a human being but a very difficult one for a computer. The first thing we need to do, since computers are only good at working with 1s and 0s, is to transform the image into a numerical matrix using pixels. As an example, Figure 5.5 shows a black and white image for a single digit number, 8, represented by a 28x28 pixel matrix. While human beings can easily recognize the image as a number 8 by some magic sensors in our eyes, a computer needs to input all of the 28x28=784 pixels, each having a pixel value—a single number representing the brightness of the pixel. The pixel value has possible values from 0 to 255, with 0 as black and 255 as white. Values in between make up the different shades of gray. If we have a color image, the pixel...