Deep Learning

 Flatten can convert 2D image to a 1D tensor Example-Image - [ 3 1 5   1 1 5] Flatten- [ 3 1 5 1 1 5 ]

 This layer is used to downsample the shape of an image Suppose you have an image- [ 2 3 4 5   3 4 5 6   4 5 6 7 ] if pool size is (2,2) then output will [ 4 6   5 7 ]

This is the main layer of feature extraction part of the convolution neural networks. Convolution layer They had a filter of shape (x,x), generally x=3 so shape of filter is (3,3). Kernel 2 2 1 3 3 3 3 2 1 Image 2 3 5 3 3 3  (2*2)+(3*2)+(5*1)+(3*3)+(3*3)+(3*3)+(3*3)+(2*2)+(1*1)=56 3 2 1 Every pixel will change by 56 of image