-- Nov 15 In-Class Exercise
We are using the identity function as our activation.
Our filter is given by
` [[1,0], [0,1]] `
Input for first neuron is
`[[1,0], [0,1]]`
So, output of neuron 1 of first image is dot product of this input with filter, equal to 2.
Input for second neuron is
` [[0,0], [1,0]]`
output of neuron 2 of first image is dot product of this input with filter, equal to 0.
Doing like this, for each of the neuron, we get, the feature maps as:
` f1 = [[2,0,1,0], [0,2,0,1], [1,0,2,0], [0,1,0,2]]`
` f2 = [[0,1,1,0], [1,1,2,1], [1,2,1,1], [0,1,1,0]]`
There are 16 neurons in each feature map.
(
Edited: 2017-11-15)
We are using the identity function as our activation.
Our filter is given by
@BT@ [[1,0], [0,1]] @BT@
Input for first neuron is
@BT@[[1,0], [0,1]]@BT@
So, output of neuron 1 of first image is dot product of this input with filter, equal to 2.
Input for second neuron is
@BT@ [[0,0], [1,0]]@BT@
output of neuron 2 of first image is dot product of this input with filter, equal to 0.
Doing like this, for each of the neuron, we get, the feature maps as:
@BT@ f1 = [[2,0,1,0], [0,2,0,1], [1,0,2,0], [0,1,0,2]]@BT@
@BT@ f2 = [[0,1,1,0], [1,1,2,1], [1,2,1,1], [0,1,1,0]]@BT@
There are 16 neurons in each feature map.