How to implement wide one dimensional convolution in cntk

Question

I'd like to implement the wide type of one dimensional convolution (https://arxiv.org/pdf/1404.2188v1.pdf) in CNTK. Is there a built in method for that or how should I play with the parameters of Convolution() to implement it?

Thanks!

Could you not specify your strides as 4,1 as opposed to like 2,2? — David Crook, Jan 11 '17 at 13:42

score 1 · Accepted Answer · answered Jan 11 '17 at 02:32

1

You could try padding your data with zero and then use C.ops.convolution().

answered Jan 11 '17 at 02:32

Sayan Pathak

870
4
7

Thanks! I think this is should be the answer for 12 filters of width 3, stride =1 in Python: Convolution((1,3), 12, init=glorot_uniform(), activation=relu, pad=True)(input) – Saghar Hosseini Jan 11 '17 at 18:26

How to implement wide one dimensional convolution in cntk

1 Answers1