Artificial Intelligence By Example

Adding a 2D convolution 

In the example of a drawing (image), we used our inbuilt human spacial module, working on a two-dimensional piece of paper with (x, y) coordinates. Two-dimensional relationships can be real-life images and also many other objects as described in this chapter. This chapter describes a two-dimensional network, although others exist:

• A one-dimensional CNN mostly describes a temporal mode, for example, a sequence of sounds (phonemes = parts of words), words, numbers and any other type of sequence
• A volumetric module is a 3D convolution, such as recognizing a cube or a video

In this chapter, a spatial 2D convolution module will be applied to images of different kinds. The main program, CNN_STRATEGY_MODEL.py, will describe ...