Group Conv Output

Problem

Apply a grouped convolution in valid mode.

The C input channels are split into groups; one shared k x k kernel per group is cross-correlated with every channel of that group and the responses are summed, producing one output channel per group. Return an (H - k + 1) x (W - k + 1) x groups map.

Test cases

case 1

in: [[[[1,2],[3,4]],[[5,6],[7,8]],[[1,1],[1,1]],[[2,2],[2,2]]],[[[1,0],[0,0]],[[0,0],[0,1]]],2]

expected: [[[1,8]],[[5,1]],[[1,2]]]

case 2

in: [[[[1,1,1],[1,1,1],[1,1,1]],[[1,1,1],[1,1,1],[1,1,1]]],[[[1,1],[1,1]]],1]

expected: [[[12],[12]]]

case 3

in: [[[[1,5],[2,6]],[[3,7],[4,8]]],[[[1,1],[1,1]],[[1,1],[1,1]]],2]

expected: [[[10,26]]]

Starter codePython

def group_conv(feature, kernels, groups):
    # Your code here
    pass

3 test cases · run your code against them in the editor