In deep learning, given 1) an input image represented by a (3x28x28) tensor, 2) 12 (3x4x4) !lters, 3) a valid, no padded, convolution, is the output: a) a (12x25x25) cube of feature maps where each value in each map is the result of the convolution of 48 values into one? b) a (12x25x25) cube of feature maps where each value in each map is the result of the convolution of 16 values into one? or c) a (12x28x28) cube of feature maps where each value in each map is the result of the convolution of 24 values into one?
In deep learning, given 1) an input image represented by a (3x28x28) tensor, 2) 12 (3x4x4) !lters, 3) a valid, no padded, convolution, is the output:
a) a (12x25x25) cube of feature maps where each value in each map is the result of the convolution of 48 values into one?
b) a (12x25x25) cube of feature maps where each value in each map is the result of the convolution of 16 values into one? or
c) a (12x28x28) cube of feature maps where each value in each map is the result of the convolution of 24 values into one?
The output is a (12x25x25) cube of feature maps, where each value in each map is the result of the convolution of 48 values into one, from an input image represented by a (3x28x28) tensor, 12 (3x4x4) filters, and a valid, no padding, convolution.
Step by step
Solved in 2 steps
