# Anyone has implemented Kernel SVD in this ICLR 2019 paper?

**URL:** <https://discourse.julialang.org/t/anyone-has-implemented-kernel-svd-in-this-iclr-2019-paper/56780>\
**Category:** Machine Learning\
**Tags:** machine-learning, network-compression\
**Created:** [March 9, 2021, 2:52am UTC](https://discourse.julialang.org/t/anyone-has-implemented-kernel-svd-in-this-iclr-2019-paper/56780 "2021-03-09T02:52:15Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Lian\_Yunlong](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/lian_yunlong/32/14437_2.png) [@Lian\_Yunlong](https://discourse.julialang.org/u/Lian_Yunlong)\
**Post date:** [March 9, 2021, 2:52am UTC](https://discourse.julialang.org/t/anyone-has-implemented-kernel-svd-in-this-iclr-2019-paper/56780/1 "2021-03-09T02:52:15Z")

</div>

Hello experts,  
recently I am trying out `Flux.jl` to initiate a research project on ML. I quickly ran into the issue of network compression and knowledge distillation. I have a background in physics and I have heard about the success of SVD decomposition in tensor network algorithms. Then I came across this paper

The Singular Values of Convolutional Layers  
(by Hanie Sedghi, Vineet Gupta and Philip M. Long from Google Brain)

> **[The Singular Values of Convolutional Layers](https://arxiv.org/abs/1805.10408)**
>
> We characterize the singular values of the linear transformation associated with a standard 2D multi-channel convolutional layer, enabling their efficient computation. This characterization also leads to an algorithm for projecting a convolutional...

My question is :  
has anyone implemented the Flux equivalent of the kernel SVD in this paper ?

Thanks in advance 😃