Conv1D on 2D input

Question

Can someone explain to me what happens when a keras Conv1D layer is fed 2D input? Such as:

model=Sequential()
model.add(Conv1D(input_shape=(9000,2),kernel_size=200,strides=1,filters=20))

Varying the input size between (9000,1) and (9000,2) and calling model.summary(), I see that the output shape stays the same, but the number of parameters changes. So, does that mean that different filters are trained for each channel, but the output is summed/averaged across the 2nd dimension before outputting? Or what?

Answer 1

Here is a visual illustration

kernel_size = (2, )

------------- 
| 1 1 1 1 1 |                <---- kernel dim = kernel_size X 5
| 2 2 2 2 2 |
-------------
  3 3 3 3 3 



--------------------------
| 1 1 1 1 1 1 1 1 1 1 1 1 |  <---- kernel dim = kernel_length X 12
| 2 2 2 2 2 2 2 2 2 2 2 2 |        i.e more params! but after 
--------------------------         you apply say MaxPool1D(pool_size=(2,2))
  3 3 3 3 3 3 3 3 3 3 3 3          in both cases, then layer shapes from here
                                   on out are the same, thus same outputs!