Calculate the output feature map size after a max pooling operation.
Max pooling downsamples feature maps by taking the maximum value in each pooling window. The output size is computed as floor((input + 2×padding − pool_size) / stride) + 1. Standard 2×2 max pooling with stride 2 halves the spatial dimensions.
Pooling Output
output = floor((input + 2*padding - pool_size) / stride) + 1
No — max pooling is a fixed operation with no learned parameters. It simply selects the maximum value in each window.
Max pooling preserves the strongest feature activations and provides some translation invariance. Average pooling is smoother and preferred in later layers or for global aggregation.