Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- Mathematically, let's say you have:
- - Input image or feature map from the previous layer: I (with dimensions Height x Width x Number of Channels).
- - Filters: F (with dimensions Filter Height x Filter Width x Number of Input Channels x Number of Output Channels).
- For each position (i, j) in the output feature map (let's call it O) of the second convolutional layer:
- - Take a receptive field from the previous layer (I) that corresponds to the size of the filter. This receptive field is centered around the position (i, j).
- - Perform an element-wise multiplication between the filter weights (F) and the values in the receptive field (I) at that position.
- - Sum up the results of the element-wise multiplication to get a single value at position (i, j) in the output feature map.
- Mathematically, this can be represented as:
- O(i, j, k) = Σ Σ Σ Σ ( I(a, b, c) * F(p, q, c, k) )
- Where:
- O(i, j, k) is the value in the output feature map at position (i, j) for the k-th filter.
- I(a, b, c) is the value in the input feature map at position (a, b) in channel c.
- F(p, q, c, k) is the weight of the filter at position (p, q) for input channel c and output channel k.
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement