Strides determine how a sliding window moves over the input data during a convolution or pooling operation. For common cases, the strides are [1, stride, stride, 1] where the stride is the same for width and height. Padding controls how the input and output sizes are calculated, with "SAME" padding matching output size to input size by padding edges and "VALID" padding using only valid regions with no padding.

1. Strides & Padding
D.COM
DEEP LEARNING STUDY
BAEK

2. Strides
Must have strides[0] = strides[3] = 1.
For the most common case of the same horizontal and vertices strides, strides = [1, stride, stride, 1].
•strides: A list of ints. 1-D tensor of length 4.
•The stride of the sliding window for each dimension of input.
•The dimension order is determined by the value of data_format

3. Strides
Strides =
[1,1,1,1]

4. Padding

5. Padding
5x5 > 32x32 to 28x28
5x5 > 36x36 to 32x32

6. Padding
If padding == "SAME":
output_spatial_shape[i] =
ceil(input_spatial_shape[i] / strides[i])
If padding == "VALID":
output_spatial_shape[i] =
ceil((input_spatial_shape[i] - (spatial_filter_shape[i]-1) * dilation_rate[i]) / strides[i]).

7. Pooling [ Max Pooling]
Down
sampling