Asymmetric Padding support in MXNet

[ChaiBapchya]

Problem

Feature request for "Asymmetric padding"
Currently, MXNet supports Symmetric padding
e.g.

Input data -> 1,2,3,4 and pad=2
Output data -> 0,0,1,2,3,4,0,0

Why isn't Asymmetric padding supported by CUDnn, MXNet?

Due to misalignment of input-output if padding happens only on one side (left or right alone).
It is called as "Causal Convolution" (only one side pad)

Current work-around

Level of abstraction introduced.
By default, one can use pad parameter while performing Pooling or Convolution which performs symmetric padding. But, if there is a specific use-case for performing asymmetric padding, one has to perform following steps,
Perform independent padding (using pad operator) and then perform Pooling or Convolution as desired.

Future steps

If one has to fully implement asymmetric padding, following changes need to be made

Pooling

Add another parameter such as pad_type (because currently, pooling_conv which allows - 'full', 'valid', 'same' has been misused. Original meaning of pooling_conv was not supposed to take care of "type of padding"

Convolution

Incorporate asymmetric padding for the same

Reference

PR #12594 that supports MaxPool 1D with "same" padding

[ChaiBapchya]

@mxnet-label-bot add [Operator, Feature request]

[marcoabreu]

I think we had this request due to onnx and it got implemented somewhere

@nswamy (don't have roshanis handle memorized)

[apeforest]

@Roshrini Do you know if this feature already exists for onnx?

[Roshrini]

@apeforest No this feature does not exist yet. We use the work around mentioned by Chaintanya (Using pad op with pooling)