Terms of Deep Learning

(*) Zero Padding -- Adding zero at the edges

Why we need zero padding?
Ans: If there is information at the edge of the image we will lost those information while applying filter on the image. To overcome this problem we need to add zero padding.

(*) Stride -- Stride means step to move the filter across the image.

(*) Depth -- Depth means how man filters are applied in the image. 3 filters are applied means the depth is 3.

(*) Max Pooling -- put the maximum value of the focused area. Stronger pixels are retained, not the lowest one.

	After each convolution layer we use max pooling. It reduces the dimension of input image that is being sent to the next layer.

This reduces the computational load and over fitting.
(*) Average Pooling -- put the average value of the focused area.

It fails to detect the sharp edges and complex features.

(*) Sum Pooling -- put the sum of all values of the focused area.





Convolutional Layer > Pooling layer > Flattening Layer > The first full connected layer > Fully connected output layer




--------------------------------
Lets build a CNN Model:
--------------------------------