C1 LayerΒΆ

The C1 layer is the first complex cell layer. It computes a MAX function over S1 inputs. The C1 layer consists of grids that are each defined by a <grid> object, as follows:

<grid>
   name c1_48         # This name is made up by us
   xn 4               # Number of units along the x-axis
   dx 24       (pix)  # X-offset between units
   width 48    (pix)  # Overall C1 RF size, determines how many S1 inputs
   tiles 1 2          # Names of S1 RF scales that are integrated
</grid>

The first parameter assigns a name to the grid. The next, “xn” defines the width (and height) of the grid in terms of numbers of units. The C1 units within the grid will be displaced horizontally (and vertically) by “dx” pixels. The “width” specifies the width (and height) in S1 units of the square region of S1 units that will be pooled by the MAX function. The “tiles” parameter specifies which S1 tile channels will be pooled by the MAX function. The value can be a list, and in this case specifies the tiles named “1” and “2”.